the return on investment of pandemic rental …

THE RETURN ON INVESTMENT OF PANDEMIC RENTAL

ASSISTANCE: MODELING A RARE WIN-WIN-WIN*

SAM GILMAN**

ABSTRACT

We are facing an eviction crisis. The COVID-19 pandemic has sent oureconomy into a tailspin, forcing countless Americans to choose between feedingtheir families or having a roof over their heads. Many low-income people,especially low-income people of color, are facing an unprecedented economiccrisis with tremendous rates of wage reductions and job loss. This has resulted inmillions of Americans being unable to pay their full rents, creating the legalgrounds for their landlords to evict them.1 Scholars, policymakers, and advocateshave increasingly focused on a number of solutions to the eviction crisis(including eviction moratoria and rental assistance), concluding that thesesolutions can stabilize households, especially when combined. Yet, a counternarrative almost always offers that investing in national rental assistanceprograms will be expensive. Indeed, advocates have estimated that more than$100 billion in funding may be needed.2 However, few analysts have emphasizedthe financial costs of inaction. This Article aims to fill that gap by estimating theReturn on Investment (“ROI”) of pandemic-related rental assistance programs by

* Copyright © 2021, Sam Gilman. All Rights Reserved. This Article requires an enormous

number of thank yous to the village of people who contributed. I owe deep gratitude to Jac Woo.

She painstakingly checked my eviction risk model. She provided heroic efforts to strengthen the

model, account for differentials in job and wage losses, and help evaluate rental debt shortfalls. She

also helped me problem solve the cost-benefit analysis model. I also owe huge thank yous to Steven

Andrew Reid and Sam Shaffer for respectively providing help on critical components of this model.

Thanks to Zach Neumann and the rest of the COVID-19 Eviction Defense Project team for the push

to continue to produce research to shape the response. Neil Steinkamp inspired me to take on this

project and has helped me shape and frame this analysis from the macro to the micro in weekly

check-ins since the beginning of the project. Mark Fagan created an opportunity for me to work on

this project through an independent study at the Harvard Kennedy School and has guided my

writing process. Thank you to Nicole Summers for giving the piece a critical read. Thank you to

Nicole Summers, Esme Caramello, and Eloise Lawrence for teaching me housing law and

informing the essential eviction framing. Finally, I am grateful to the Indiana Health Law Review

and its editors for their thoughtful edits and endless patience.

** Sam Gilman works at the intersection of law, policy, and systems. He is a co-founder and

COO at the COVID-19 Eviction Defense Project and a JD/MPP student at Harvard University.

1. Tabulations of Week 20 of the Household Pulse Survey, COVID-19 EVICTION DEF.

PROJECT (Dec. 16, 2020), https://cedproject.org/2020/12/18/tabulations-of-week-20-of-hh-pulse-

survey/ [https://perma.cc/NLU3-VSM3].

2. ANDREW AURAND ET AL., NAT’L LOW INCOME HOUS. COAL., NLIHC RESEARCH NOTE:

EMERGENCY RENTAL ASSISTANCE NEEDS FOR WORKERS STRUGGLING DUE TO COVID-19 3 (2020),

https://nlihc.org/sites/default/files/Emergency-Rental-Assistance-Needs-for-Workers-Struggling-

due-to-COVID-19.pdf [https://perma.cc/JHP3-YLV6].

294 INDIANA HEALTH LAW REVIEW [Vol. 18:293

comparing the costs of rental assistance with the social costs of homelessness anddisplacement. As seen in Figure 1, and using end-of-2020 source data, thisArticle estimates that pandemic rental assistance has a positive ROI ofbetween 208-466%. These ROI values point to the conclusion that failing toinvest in rental assistance will cost dramatically more than making the investmentnow. The ROI analysis finds that rental assistance stabilizes both tenants andlandlords, preserves neighborhoods, and protects government budgets over thelong-term. More broadly, the returns on rental assistance argue for a re-imagination of the eviction system. The conclusion that the estimated benefits ofrental assistance eclipse the estimated costs of providing the funds by three orfour times suggests that rental assistance should supplant eviction as the socialremedy for the inability to pay rent. With an investment of rental assistance, non-payment evictions can forever remain “non-essential” evictions. In short, keepingpeople in their homes during this pandemic and beyond is not only the right thingto, but it is economically the smart thing to do.

Figure 1

I. INTRODUCTION

A. Background: A Financial Crises Triggers and Eviction Crisis

The COVID-19 pandemic has sparked one of the most severe human andeconomic crises of our lifetime. Ten months into the pandemic, the United Stateseconomy had ten million fewer jobs—a level surpassing that of the GreatRecession.3 The pandemic’s affect has demonstrated the harsh economic reality

3. Jared Bernstein & Heather Boushey, The Employment Situation in January, WHITE

HOUSE (Feb. 5, 2021), https://www.whitehouse.gov/briefing-room/blog/2021/02/05/the-

2021] THE RETURN ON INVESTMENT OFPANDEMIC RENTAL ASSISTANCE

295

facing low-income households who have unemployment rates nearing 20%.4

Black and brown people have been affected financially at significantly higherrates than white people, exacerbating a climate of income inequality.5 And whilemany men have returned to work, women – and primarily women of color – havenot.6

These economic conditions have triggered a housing crisis of historicportions, on top of a preexisting housing crisis.7 Tens of millions of Americansare now at risk of eviction and displacement. As of mid-December, nearlynineteen million individuals lived in households behind on rent and arrearages areaccumulating with each passing month.8

Missing a single rent payment in America, particularly during a globalpandemic, can trigger a property owner’s right to evict non-paying renters fromtheir homes. In general, if someone gets sick, their car breaks down, or theiremployer closes their doors, they risk losing their home. This creates a financialspiral that can result in a swift eviction and descent into poverty andhomelessness. As Matthew Desmond writes: an eviction is not merely aconsequence of poverty – it is often a cause of poverty.9 Evictions have beenlinked to job loss, difficulty finding future housing, homelessness, chronic illness,poor learning outcomes, generational poverty, diseases of despair, and now deathby COVID-19.10

It is not just the tenants who lose when they miss rent. Landlords experiencesteep financial losses – more so during the pandemic. According to the NationalApartment Association, landlords who rent affordable housing often collect lessthan 10% of rental debts.11 And in the end, government pays for many of the

employment-situation-in-January/ [https://perma.cc/ZW9R-B8R6].

4. See Raj Chetty et al., The Economic Impacts of COVID-19: Evidence from a New Public

Database Using Private Sector Data 3, 25 (Nat’l Bureau of Econ. Research, Working Paper No.

27,431).

5. Bernstein & Boushey, supra note 3.

6. Claire Ewing-Nelson, Another 275,000 Women Left the Labor Force in January, NAT’L

WOMEN’S L. CTR. (Feb. 2021), https://nwlc.org/wp-content/uploads/2021/02/January-Jobs-Day-

FS.pdf [https://perma.cc/SA24-ELVE].

7. See infra Part II.B.

8. See Week 21 Household Pulse Survey: December 9 – December 21, U.S. CENSUS

BUREAU, https://www.census.gov/data/tables/2020/demo/hhp/hhp21.html [https://perma.cc/F6G6-

YKQC] (last updated Jan. 5, 2021).

9. See, e.g., MATTHEW DESMOND, EVICTED: POVERTY AND PROFIT IN THE AMERICAN CITY

(2016).

10. See infra Part III (discussing the various financial and physical health consequences of

an eviction); see also infra Part VII.E (detailing the research into health and welfare consequences

of shelter-based homelessness and non-shelter-based displacement).

11. Multifamily Debt Collections: Best Practices for Owners and Managers, NAT’L

APARTMENT ASS’N (Sept. 2019), https://www.naahq.org/news-publications/units/september-

2019/article/multifamily-debt-collections-best-practices [https://perma.cc/GG38-6H4V].


downstream consequences of eviction, including housing and services for theunhoused, health care for low-income individuals, unemployment benefits, fostercare for children whose parents do not have a safe home, foreclosure when alandlord cannot sustain their home, and many other costs.

B. Asking the Financial Question

By January 2021, Americans had accrued an estimated $24-$57 billion inrental and utility debts.12 By June, those debts are estimated to climb toapproximately $45-60 billion, accounting for federal rental assistance alone.13 Bythe end of the year, shortfalls for rental assistance could reach over $120 billion.

From a human perspective, it is clear that a rental assistance program thatpays peoples’ rents and avoids eviction would be worth the investment, wardingoff the life-altering and life-threatening consequences of displacement, eviction,and homelessness. It would also make landlords whole or close to it.

But from a financial perspective, the question is: how much would a rentalassistance safety net cost? And would it pay for itself? If the answer to the latterquestion is yes, advocates can make a compelling case for a comprehensiveeviction prevention and rental assistance program that ensures no household isevicted for non-payment of rent or without “good” or “just” cause during thepandemic and recovery.14 Under this regime, we can retain the standard identifiedin Massachusetts’ eviction moratorium, defining “non-payment of rent,” no-faultevictions, and for cause evictions that do not threaten health and safety as “non-essential eviction[s].”15

This Article was motivated by an effort to answer that financial question. Itprovides a directional estimated ROI for three rental assistance programs. Itevaluates: (1) a program structured similarly to the Emergency Rental AssistanceProgram in the Consolidated Appropriations Act; (2) a more expansive federalrental assistance program; and (3) the expansive program including partial rentaldebt settlements.

C. Methodology

This Article presents a directional cost-benefit analysis that estimates thesocial value of each dollar invested in rental assistance, closely mirroring themethodology of Stout Risius Ross’s (“Stout”) cost-benefit-analyses of right-to-counsel efforts in New York, Baltimore, Los Angeles, and Philadelphia.16 In each

12. See infra Part II.C (referencing three estimates for rental debts, including Moody's

Analytics; Stout Risius Ross; and this model from the COVID-19 Eviction Defense Project).

13. See id.

14. “Just Cause” statutes limit the grounds upon which landlords can evict tenants, often

including property damage, major lease violations, and non-payment. See, e.g., N.J. STAT. ANN.

§ 2A:18-61.1 (West 2021); D.C. CODE § 42-3505.01 (2021).

15. See 2020 Mass. Acts 65.

16. See generally STOUT RISIUS ROSS, INC., THE FINANCIAL COST AND BENEFITS OF

ESTABLISHING A RIGHT TO COUNSEL IN EVICTION PROCEEDINGS UNDER INTRO 214-A (2016),


297

geography, these studies evaluate the relative benefits of right-to-counselprograms, meaning a right to legal representation in eviction cases.17

Following the Stout approach, this Article estimates the number of formal andinformal evictions reduced by rental assistance programs. For each modeledprogram, it multiplied these estimates of eviction risk by (a) the estimated per-individual cost of the rental assistance program and (b) compared these costs tothose that government and society would not pay due to the avoided eviction.This cost-benefit analysis establishes a directional quantified ROI that includesestimates of the public costs and social costs that are avoided due to theintervention. Public costs are those that governments pay as a result of eviction.Social cost estimates supplement public costs with additional private costs thataccrue to society. For example, the total social cost includes the public costs aswell as broader neighborhood, landlord, and employment costs.

It is worth noting that the ROI modeling has several important limitations.First, the analysis is built-off of estimates of displacement risk and economicrecovery during the pandemic where there are few good data sources and evenfewer historical analogs. Second, the model uses data from a period in whichlimited financial assistance was available to struggling families, but it does nottake into account the interaction between rental assistance and other interventionslike cash payments. Third, this analysis is severely limited by insufficient data onthe outcomes of eviction and displacement. It, therefore, is likely to understate thecosts avoided by evictions that did not happen. Fourth, fifty-state data is subjectto enormous national variability and fluctuations in weekly Household PulseSurvey data.18

These limitations point to the need for future experimental and empiricalresearch on the impacts of rental assistance. Future studies should applyexperimental methods to understand the effects of rental assistance on eviction,displacement, shelter use, rehousing, and all of the costs associated with each ofthose outcomes in different geographies. This Article aims to provide an outline

https://cdn2.hubspot.net/hubfs/4408380/PDF/Cost-Benefit-Impact-Studies/SRR%20Report%20-

%20Eviction%20Right%20to%20Counsel%20%203%2016%2016.pdf [https://perma.cc/456D-

4GLH].

17. See generally id. The authors estimate the ROI of the right-to-counsel in each respective

city by comparing the costs of the program with the public costs avoided due to incremental

reductions in evictions. These studies have three primary building blocks: (1) the number of

evictions reduced by legal representation, which sets the multiplier, (2) the all-in costs of providing

attorneys, and (3) the public costs avoided from reducing evictions. Social benefit measures are

calculated by estimating the cost of a given intervention and accounting only for the portion of the

costs avoided that can be attributable to the intervention according to academic studies.

18. It is further limited in a number of instances where no fifty-state analysis exists. For

example, in some cities and states, the convention is for renters seeking new housing to have to pay

first-months, last-months and security deposit. In others, renters must only pay first months’ rent

and security deposit. This type of variability could not be accounted for. Conservative assumptions

were chosen to limit the risk of overestimate.


of the scope of such an analysis and makes the case that experimental data on therole of rental assistance is essential to making the definitive case that rentalassistance programs are high-ROI interventions.

D. Roadmap

The remainder of this Article consists of six parts. Part II quantifies theeviction and rental debt crisis, synthesizing available research and presenting newnumbers from the Aspen Institute-CEDP Eviction Risk Model. Part III outlinesthe consequences of evictions for renters, landlords, and the broader housingmarket. Part IV details the proposed solutions to the housing crisis, includingeviction moratoria and rental assistance. Part V presents the quantified ROIanalysis for federal rental assistance. Part VI sets forth a reimagination of theeviction-industrial complex. It posits that housing stabilization through rentalassistance should supplant eviction as the primary social remedy for non-paymentof rent. While landlords should retain the right to evict for “just cause” and evenchronic non-payment of rent (where rental assistance cannot sustain the tenancy),this research suggests that all parties would be better off with a strengthenedhousing safety net. Such policies would more appropriately balance tenants’ rightto housing and landlords’ ability to earn income from their properties. Part VIIpresents a detailed methodological appendix.

II. QUANTIFYING THE EVICTION AND RENTAL DEBT CRISIS

A. Background: Preexisting Housing Crisis

The COVID-19 virus attacked an American rental economy that was alreadyenduring a deep housing affordability crisis. Nearly half of American renters werecost burdened, paying more than 30% of their incomes on rent in 2019.19 Aquarter of renters were severely cost burdened, paying more than 50% of theirincomes on rent.20 More than half a million Americans experiencedhomelessness—either living on the streets or in emergency shelters.21 Accordingto the National Low Income Housing Coalition’s Out of Reach report, “modestrental housing is out of reach for every worker in the bottom half of the wagedistribution.”22 It also has a disproportionate effect on renters of color.23

19. JOINT CTR. FOR HOUS. STUDIES OF HARVARD UNIV., AMERICA’S RENTAL HOUSING 2020,

at 5 (2020), https://www.jchs.harvard.edu/sites/default/files/Harvard_JCHS_Americas_Rental_

Housing_2020.pdf [https://perma.cc/8FEB-LPDX]

20. Id.

21. See U.S. DEP’T. OF HOUS. & URBAN DEV., THE 2019 ANNUAL HOMELESS ASSESSMENT

REPORT (AHAR) TO CONGRESS PART 1: POINT-IN-TIME ESTIMATES OF HOMELESSNESS 8 (2020),

h t tps:/ /www.huduser.gov/portal/sites/de fau lt / f i le s /pdf /2 0 1 9 -AHAR-Part-1 .pdf

[https://perma.cc/HTY2-42DW].

22. ANDREW AURAND ET AL., OUT OF REACH: THE HIGH COST OF HOUSING 5 (2020),

https://reports.nlihc.org/sites/default/files/oor/OOR_2020.pdf [https://perma.cc/G4MU-DXHU].

23. Chester Hartman & David Robinson, Evictions: The Hidden Housing Problem, 14


299

According to 2020 research by the Massachusetts Institute of Technology andCity Life/Vida Urbana, 70% of Boston’s market-rate evictions were filed incensus tracts where the majority of residents are people of color despiteaccounting for only half of the city’s rental housing.24

Due to the rental affordability crisis, the United States was already in themidst of a crisis of mass eviction and displacement when the pandemic hit.Studies estimate that more than 70% of evictions occur due to non-payment ofrent.25 On average, landlords file 3.6 million eviction cases per year according toThe Eviction Lab.26 A Matthew Desmond and Tracey Schollenberger study inMilwaukee found that for every formal eviction that led to forced displacementthere were an additional two households that left their homes due to informalevictions.27

B. Eviction Risk in the Pandemic

According to the Urban Institute, more than 20 million individuals live inrenter households who suffered a COVID-19 related job loss.28 The medianincome of a renter household is $40,531,29 and nearly 40% of workers makingunder $40,000 a year lost their job in the initial stages of the pandemic.30 As seen

HOUSING POL’Y DEBATE 461, 467 (2003).

24. DAVID ROBINSON & JUSTIN STEIL, EVICTIONS IN BOSTON: THE DISPROPORTIONATE

EFFECTS OF FORCED MOVES ON COMMUNITIES OF COLOR 8 (2020), https://d3n8a8pro7vhmx.

cloudfront.net/themes/5eee7e564445ea4f9a6f3080/attachments/original/1592786979/EvictionR

eport_Final_Spreads.pdf?1592786979 [https://perma.cc/B9ZB-QVXR].

25. See, e.g., Chris Salviati, Rental Insecurity: The Threat of Evictions to America’s Renters,

APARTMENT LIST (Oct. 20, 2017), https://www.apartmentlist.com/research/rental-insecurity-the-

threat-of-evictions-to-americas-renters#fn-4 [https://perma.cc/MRB7-34PZ]; see generally Robert

Collinson & Davin Reed, The Effects of Evictions on Low-Income Households (Oct. 2018)

(unpublished manuscript), https://economics.nd.edu/assets/303258/jmp_rcollinson_1_.pdf

[https://perma.cc/W4BQ-2TJP].

26. National Estimates: Eviction in America, EVICTION LAB (May 11, 2018), https://

evictionlab.org/national-estimates/ [https://perma.cc/9V2T-AHLB].

27. Matthew Desmond & Tracey Schollenberger, Forced Displacement from Rental Housing:

Prevalence and Neighborhood Consequences, 52 DEMOGRAPHY 1751, 1761 (2015).

28. See SARAH STROCHAK ET AL., HOW MUCH ASSISTANCE IS NEEDED TO SUPPORT RENTERS

THROUGH THE COVID-19 CRISIS? (2020), https://www.urban.org/sites/default/files/publication/

102389/how-much-assistance-is-needed-to-support-renters_1_1.pdf [https://perma.cc/5AJA-

KXH9].

29. Median Household Income the Past 12 Months (In 2018 Inflation-Adjusted Dollars) By

Tenure, U.S. CENSUS BUREAU, https://data.census.gov/cedsci/table?y=2018&tid=ACSDT1Y2018.

B25119&t=Income%20and%20Poverty%3AOwner%2FRenter%20%28Tenure%29&hidePrevi

ew=true&vintage=2018&g=0100000US&moe=false [https://perma.cc/DJ55-9TPL] (last visited

Feb. 25, 2021).

30. See also BD. OF GOVERNORS OF THE FED. RESERVE SYS., REPORT ON THE ECONOMIC


in Figure 2, data from Harvard’s Opportunity Insights Recovery Tracker programindicates that the recovery has been painfully slow for the bottom quartile ofrenters. As of this writing, employment rates were 21% below pre-pandemiclevels for the bottom quartile.31 Research shows that the same households thatwere hard hit by the pandemic were already suffering from housing insecurity.While some of the economy has recovered, the recovery has been slowest toreach low-income renters who disproportionately work in low-wage and serviceindustry jobs.32

Figure 2: Percent Change in Employment by Wage Quartile33

WELL-BEING OF U.S. HOUSEHOLDS IN 2019, FEATURING SUPPLEMENTAL DATA FROM APRIL 2020,

at 53 (2020), https://www.federalreserve.gov/publications/files/2019-report-economic-well-being-

us-households-202005.pdf [https://perma.cc/D4YF-6GCR]; see also Full Transcript: Fed Chair

Jerome Powell’s 60 Minutes Interview on Economic Recovery from the Coronavirus Pandemic,

CBS NEWS (May 17, 2020), https://www.cbsnews.com/news/full-transcript-fed-chair-jerome-

powell-60-minutes -in t e rv iew -econ om ic-recovery-f rom-coronavirus-pandemic/

[https://perma.cc/GZD7-PVAA].

31. See Chetty et al., supra note 4.

32. Jeremy Hobson, Economist Joseph Stiglitz: Recession Will Hurt Low-Income People

More Than the Wealthy, WBUR (Oct. 9, 2020), https://www.wbur.org/hereandnow/2020/10/09/

joseph-stiglitz-economic-recovery [https://perma.cc/7Y6H-8ZST].

33. OPPORTUNITY INSIGHTS ECON. TRACKER, https://tracktherecovery.org/ [https://perma.cc/

EJN6-ZEBS] (last visited Feb. 26, 2021).


301

This economic crisis has placed tens of millions at risk of eviction. LastAugust, an Aspen Institute analysis leveraged Census Bureau data and estimatedthat between thirty to forty million Americans could be at risk of eviction by theend of 2020, as seen in Figure 3.34 This estimate drew on two independentanalyses, one by this author and a team at Aspen and the COVID-19 EvictionDefense Project35 and another by Stout.36

Figure 3

As the pandemic surged, a patchwork of eviction moratoria protected millionsof renter households from eviction nationwide during 2020.37 In addition to forty-three state eviction orders, federal eviction protections had the effect ofdepressing eviction filings during the summer and keeping tenants in theirhomes.38 By the end of the summer, when protections began to lapse, evictionfilings reached up to four times their historical weekly averages according to datafrom The Eviction Lab.39

34. Emily Benfer et al., The COVID-19 Eviction Crisis: An Estimated 30-40 Million People

in America Are at Risk, ASPEN INST. (Aug. 7, 2020), https://www.aspeninstitute.org/blog-posts/the-

covid-19-eviction-crisis-an-estimated-30-40-million-people-in -america-are-at-r isk/

[https://perma.cc/9A24-7CEW] [hereinafter The COVID-19 Eviction Crisis].

35. Katherine Lucas McKay et al., National Eviction Risk Projections, ASPEN INST. (Aug.

10, 2020), https://www.aspeninstitute.org/publications/national-eviction-risk-projections/ [https://

perma.cc/7Z2P-YKAT].

36. See The COVID-19 Eviction Crisis, supra note 34.

37. Emily Benfer et al., Eviction, Health Inequity, and the Spread of COVID-19: Housing

Policy as a Primary Pandemic Mitigation Strategy, 98 J. URB. HEALTH 1, 11 (2021).

38. Id.

39. Eviction Tracking System, EVICTION LAB, https://evictionlab.org/eviction-tracking


In September, the Federal Government again stepped in to prevent someevictions. The Centers for Disease Control and Prevention (“CDC”) issued amoratorium on evictions for renters who attest to economic hardship duringCOVID-19, meet certain income qualifications, and affirmatively declare theirright to protection reprieve from eviction.40 A number of other states had issuedmore comprehensive eviction protections that have supplemented the CDCmoratorium.41

Additionally, the federal government, states, and localities allocated hundredsof millions of dollars for rental assistance in 2020. The National Low IncomeHousing Coalition estimates that approximately $4 billion was allocated towardsrental assistance in 2020 with the majority coming from the Coronavirus ReliefFund.42

These policies had significant, but incomplete effects. For example, the CDCmoratorium had a strong effect initially, but its effectiveness waned. After theCDC order, filings dropped below weekly averages, bottoming out at 83% ofhistorical averages.43 In Massachusetts, which formerly had the strongestmoratorium in the nation, eviction filing rates returned to pre-pandemic levels inNovember.44 Overall, The Eviction Lab’s tracker illustrates the fact that evictionsand displacements have been slowed, but not stopped by these protections withwide variability by state.45 As of mid-February 2021, nearly 250,000 evictionshad been filed in the five states and twenty-seven cities tracked by the EvictionLab.46 And, eviction risk prevails despite the existence of moratoria, as increasingnumbers of renters are behind on rent.

The August projections were not far off.47 In other words, as seen in Figure4, 18% of renters, nineteen million individuals in eight million households, wereat risk of eviction for non-payment of rent in January. Up to thirty-three millionAmericans in fourteen million households had low confidence in their ability topay. Specifically, housing insecurity disproportionately impacts Black and brown

[https://perma.cc/5HL4-6VMU] (last visited Feb. 26, 2021).

40. See Temporary Halt in Residential Evictions to Prevent the Further Spread of COVID-19,

85 Fed. Reg. 55,292 (Sept. 4, 2020).

41. See, e.g., Colo. Exec. Order No. D 2020 227 (Oct. 21, 2020).

42. REBECCA YAE ET AL., NAT’L LOW INCOME HOUS. COAL., NLIHC RESEARCH NOTE:

EMERGENCY RENTAL ASSISTANCE PROGRAMS IN RESPONSE TO COVID-19 1, 5 (2020),

ht tps:/ /n lihc.org/sites/defau lt /f iles/Emergency-Rental-Assistance-Programs-3.pdf

[https://perma.cc/8MCY-QDDN].

43. See Eviction Tracking System, supra note 39.

44. Mass. Trial Court Dep’t of Research & Planning, All Residential Eviction Cases, Non-

Payment of Rent, TABLEAU PUB., https://public.tableau.com/profile/drap4687#!/vizhome/

MassachusettsTrialCourtSummaryProcess/SummaryProcess [https://perma.cc/H7YN-M8CR] (last

visited Feb. 23, 2021).

45. See Eviction Tracking System, supra note 39.

46. Id.

47. The COVID-19 Eviction Crisis, supra note 34; Week 21 Household Pulse Survey:

December 9 – December 21, supra note 8.


303

households and families with children.48 As Figure 5 depicts, Black and Latinehouseholds were more than twice as likely to be behind on rent as whitehouseholds. A similar ratio holds for households with children compared tohouseholds without children.

Figure 4

Figure 5

As noted above, actual eviction filings and rates of displacement may trailand will ideally never reach risk projections, however, collectively, they paint apicture of systemic risk in the rental housing market and a staggering amount of

48. Week 21 Household Pulse Survey: December 9 – December 21, supra note 8.


potential suffering. Without further protection, in the form of eviction moratoriaand rental assistance, this eviction storm is projected to make landfall across thecountry.

C. Systemic Housing Market Risk: Rental Debts Threaten Tenants,Government, and Landlords Alike

Even if moratoria are in place, many renters are accumulating massive rentaland utility debts. Given that evictions are the legal response to non-payment ofrent, the presence of these debts puts renters at risk of eviction and landlords atrisk of foreclosure.49 The accumulation of rental and utility debt representssystemic risk in the housing market: if tenants cannot pay their rent,landlords—particularly mom-and-pop landlords—cannot pay their mortgages,increasing the prevalence of foreclosure for landlords and cascading effects inneighborhoods across the country.50

Historically, landlords moved to evict tenants when they fell behind on rent,often for arrearages of less than $600 and primarily for amounts less than$3,000.51 The pandemic has not only created conditions for more renters to fallbehind on rent, but it has also deepened rental debts. Early 2021 projections ofrental debt placed the outstanding amount between $24 billion52 and $57 billion53

by January 2021. Stout estimated that renters could accumulate between $13billion and $24 billion in rental arrears by January.54 Mark Zandi at Moody’sAnalytics estimated that, 10.2 million renters would owe $57.3 billion in rent andutilities through January for an average of $5,586 per renter.55

49. See, e.g., COLO. REV. STAT. § 13-40-101 (2021).

50. See, e.g., Whitney Airgood-Obrycki & Alexander Hermann, COVID-19 Rent Shortfalls

in Small Buildings, JOINT CTR. FOR HOUSING STUD. HARV. U. (May 26, 2020), https://www.

jchs.harvard.edu/blog/covid-19-rent-shortfalls-in-small-buildings/ [https://perma.cc/UW7R-JLU6].

51. Emily Badger, Many Renters Who Face Eviction Owe Less than $600, N.Y. TIMES (Dec.

12, 2019), https://www.nytimes.com/2019/12/12/upshot/eviction-prevention-solutions-government.

html [https://perma.cc/7FX9-69S5].

52. Estimation of Households Experiencing Rental Shortfall and Potentially Facing Eviction,

STOUT RISIUS ROSS, https://app.powerbi.com/view?r=eyJrIjoiNzRhYjg2NzAtMGE1MC00NmNjL

TllOTMtYjM2NjFmOTA4ZjMyIiwidCI6Ijc5MGJmNjk2LTE3NDYtNGE4OS1hZjI0LTc4ZGE

5Y2RhZGE2MSIsImMiOjN9 [https://perma.cc/5TKN-TRCN] (last visited Feb. 23, 2021).

53. Will Packer, Struggling Rental Market Could Usher in Next American Housing Crisis,

WALL ST. J. (Oct. 27, 2020), https://www.wsj.com/articles/struggling-rental-market-could-usher-in-

next-american-housing-crisis-11603791000 [https://perma.cc/9ANX-LSFP].

54. Id.

55. JIM PARROTT & MARK ZANDI, AVERTING AN EVICTION CRISIS 3 (2021), https://www.

moodysanalytics.com/-/media/article/2021/averting-an-eviction-crisis.pdf [https://perma.cc/ERV5-

84LU].


305

Figure 6

The COVID-19 Eviction Defense Project’s (CEDP) January model’sestimates fall in the middle of the range.56 The model projected $34 billion inrental and utility arrears by January 2020.57 The number of renters estimated tobe in arrears is up two percentage points and almost 27% from August.58

According to the Federal Reserve Bank of Philadelphia, the country has seen a70% increase in the percentage of people using credit cards to pay rent inNovember;59 42% of renters did not use regular income sources to pay fornecessities during the week prior to the survey period, including 66% of thosebehind on rent. As liquidity dried up in a worsening economy, families andfriends have less money available to support each other, landlords ran out ofresources to sustain tenancies, and the debt crisis continued to deepen.

Without further intervention, and economic recovery, rental shortfalls willcontinue to accumulate. Mark Zandi and Jim Parrott at Moody’s Analyticsestimated that rental arrears debts would reach $43 billion by June, accountingfor $25 billion in rental assistance in the Consolidated Appropriations Act of2021.60 Total arrears would be above $68 billion without those funds.

As seen in Figure 7, CEDP estimated a rental and utility shortfall of $85billion by the end of June 2021, without accounting for the emergency rental

56. See infra Part VII (describing how calculations were made).

57. Id.

58. Tabulations of Week 20 of the Household Pulse Survey, supra note 1.

59. Chris Arnold, More Americans Pay Rent on Credit Cards as Lawmakers Fail to Pass

Relief Bill, NPR (Nov. 30, 2020), https://www.npr.org/2020/11/30/938867270/more-americans-

pay-rent-on-credit-cards-as-lawmakers-fail-to-pass-relief-bill [https://perma.cc/V6X7-937E].

60. PARROTT & ZANDI, supra note 55.


assistance payments in the Consolidated Appropriations Act.61 The shortfallwould project to be $60 billion with full implementation by June. The modelprojected a shortfall of $148 billion by December 2021 or $123 billion, assumingfull implementation of Consolidated Appropriations Act rental assistancepayments. These figures exclude additional federal assistance and assume agradual economic recovery and that tenants pay the same monthly rent in 2020as 2021.62 Accounting only for tenants who make 80% area median income(“AMI”) or less, the model projects a shortfall of $61 billion in June and $105billion by the end of the year.

Figure 7

III. THE CONSEQUENCES OF AN EVICTION CRISIS

Evictions do tremendous harm to individuals and families, particularly duringthe pandemic. If there is no action to mitigate the effects of these crises, morethan twenty-five million people in more than ten million families could be at riskof trauma: extended homelessness, food insecurity, COVID-19 contraction,exacerbated health issues, lack of school,63 and even deaths of despair.64 Failureto contain the eviction risk increases the likelihood of severe COVID-19 spread,as social distancing and good hygiene are much harder for people living intemporary residences, shelters, and on the street.65

61. It is difficult to estimate rental debts into the future, especially in the absence of a federal

eviction moratoria and in the presence of an improving economy, as evictions stop the

accumulation of rental debt. For that reason, the CEDP model estimates rental shortfall – the

amount of rent that should be paid for tenants at risk of eviction in 2021 assuming no evictions.

62. See infra Part VII (discussing the difference in methodologies between Mark Zandi and

CEDP).

63. See, e.g., DESMOND, supra note 10; Matthew Desmond & Carl Gershenson, Housing and

Employment Insecurity Among the Working Poor, 63 SOC. PROBS. 46 (2016); Collinson & Reed,

supra note 25.

64. See Katherine A. Fowler et al., Increase in Suicides Associated with Home Eviction and

Foreclosure During the US Housing Crisis: Findings From 16 National Violent Death Reporting

System States, 2005–2010, 105.2 AM. J. PUB. HEALTH 311 (2015).

65. Lynn Jolicouer, Coronavirus Testing in Boston’s Homeless Community Shows Shelter

Size and Density Matter, WBUR (May 15, 2020), https://www.wbur.org/commonhealth/2020/


307

As Emily Benfer, a law professor and leading expert on evictions, and her co-authors document in a recent paper, evictions are associated with numerousphysical and mental health conditions, negative health outcomes for women(including domestic violence, negative outcomes for children), and additionalsub-standard living conditions when families move.66 And as Matthew Desmondoften highlights, evictions can often be causes in addition to consequences ofpoverty and lead to negative employment outcomes.67 They also often forcerenters to move into lower quality housing and neighborhoods with worseconditions, including crime, poverty, and declining housing stock. Additionally,evictions create a “Scarlet E” that increases the costs of housing and lowersaccess to high quality housing for tenants who are screened out based on theireviction history.68

Moreover, experimental and simulation studies have directly tied eviction toincreased incidences of COVID-19 and COVID-19 deaths. In a different paper,Benfer and her co-authors found that states that lifted moratoria during thepandemic experienced statistically significant increases in incidences of COVID-19 and mortality due to COVID-19, controlling for major state fixed effects andinterventions.69 They found that lifting moratoria accounted for hundreds ofthousands of excess cases and more than ten thousand excess deaths.70

Researchers at the National Bureau of Economic Research found, for example,that if comprehensive bans on evictions had been in place throughout thepandemic, the intervention would likely have reduced infections by 14.2% anddeaths by 40.7%.71

For mom-and-pop landlords, unpaid rent and evictions mean lost income,unpaid mortgages, and unpaid property taxes. These impacts can also lead toincome loss and similar traumatic effects for mom-and-pop landlords and theirfamilies, as with renters. According to an analysis from the Harvard Joint Centerfor Housing Studies, renters and owners of small properties are disproportionatelylikely to face COVID-19 related economic hardship.72 More than half of therenters with jobs that were at risk in the pandemic live in single-family or two to

05/15/boston-homeless-coronavirus-testing [https://perma.cc/2FT6-MFPW].

66. Benfer et al., supra note 37, at 8; see infra Part VII.

67. DESMOND, supra note 9; Desmond & Gershenson, supra note 63.

68. See The Scarlet E, WNYC STUDIOS (June 6, 2019), https://www.wnycstudios.

org/podcasts/otm/scarlet-e-unmasking-americas-eviction-crisis [https://perma.cc/7Q6Y-AV5U].

69. Kathryn M. Leifheit et al., Expiring Eviction Moratoriums and COVID-19 Incidence and

Mortality 3-4 (Nov. 30, 2020) (unpublished manuscript) (on file with the Indiana Health Law

Review)

70. Id. at 5.

71. See Kay Jowers et al., Housing Precarity & the COVID-19 Pandemic: Impacts of Utility

Disconnection and Eviction Moratoria on Infections and Deaths Across US Counties 1 (Nat’l

Bureau of Econ. Research, Working Paper No. 28,394).

72. Airgood-Obrycki & Hermann, supra note 50.


four unit homes.73 These homes represent the majority of the naturally occurringaffordable housing stock, and roughly 75% of these properties are owned bymom-and-pop landlords who rely on this rental income to pay their mortgagesand property taxes.74

Financial difficulty among small landlords could destabilize naturallyoccurring affordable housing and lead to a greater cycle of evictions. A similarcycle emerged after the Great Recession. At that time, many single-family homeswere purchased by private equity firms in foreclosure sales and turned into rentalproperties run by management companies. These corporate landlords areincentivized to maximize profits using two levers: increasing rents and fees(revenues) and limiting costs like unpaid rents and maintenance.75 Evidence fromthe last decade suggests that these landlords have become some of the mostefficient evictors and absentee landlords in the United States. A Federal ReserveBank of Atlanta Study in 2016 found that large private equity firms filedevictions on a third of their properties in the year, and they had an 18% higherhousing instability rate, controlling for property and neighborhood characteristics.It is also worth highlighting that Black residents of Atlanta were evicted at thehighest rates. In other words, private equity ownership is itself associated with ahigher housing instability rate that disproportionately impacted Black and browncommunities.76 The larger the 2020 eviction and foreclosure crisis, the moreproperties available for private purchase.

IV. PROPOSED SOLUTIONS TO THE COVID-19 EVICTION CRISIS

Eviction prevention policy primarily aims to keep people in their homes or,if necessary, provide those who must move adequate time and resources to movesafely to new homes. There are two types of policy solutions to any evictioncrisis: legal measures to prevent short- and medium-term displacement (or reducethe disruption or displacement) and cash assistance to stabilize tenancies. Ofcourse, this typology is not binary. Legal representation coupled with favorablelandlord-tenant laws can often have the effect of stabilizing households.Conversely, except for long-term rental subsidies that are effectively permanent,emergency rental assistance may merely minimize the disruption of an eviction.

Legal measures might include providing access to mediation services, accessto counsel, and process changes to lengthen the eviction timeline. Cash assistance

73. Id.

74. Id.

75. See, e.g., Michelle Conlin, Spiders, Sewage and a Flurry of Fees – The Other Side of

Renting a House from Wall Street, REUTERS (July 27, 2018), https://www.reuters.com/investigates/

special-report/usa-housing-invitation/ [https://perma.cc/T7DZ-TB7F].

76. See generally ELORA RAYMOND ET AL., FED. RESERVE BANK OF ATLANTA, CORPORATE

LANDLORDS, INSTITUTIONAL INVESTORS, AND DISPLACEMENT: EVICTION RATES IN SINGLE FAMILY

RENTALS (2016), https://www.frbatlanta.org/-/media/documents/community-development/

publications/discussion-papers/2016/04-corporate-landlords-institutional-investors-and-

displacement-2016-12-21.pdf [https://perma.cc/A87W-MLF9].


309

can come in many forms and may also include measures like the cancellation ofrent or rental debt or the conversion of rental debts into non-evictable civil debts.The former buy renters time in their homes and the latter cure rental arrears,addressing the root cause of potential evictions.

The two most prominent versions of these policies during the pandemic havebeen eviction moratoria and rental assistance given their ability to operate atscale. If implemented effectively and in tandem, eviction moratoria and rentalassistance are mutually reinforcing.77 Moratoria keep tenants in their homes fora defined period of time, and rental assistance can permanently “cure” the risk ofeviction by clearing tenant debts and paying future rents for enough time to allowtenants to stabilize. As the sections below will document, not all moratoria andrental assistance policies are created equal. Policy design has a tremendousimpact on whether the policies actually stabilize tenancies.

A. Legal Measures to Prevent Displacement

The federal government, and states in the absence of federal action, can passa number of eviction moratoria and housing security policies to buy renters andlandlords time. Four policies necessitate a special mention: eviction moratoria,foreclosure protections, civil right to counsel, and civil debt conversion.

1. Eviction Protections

The most successful moratoria prevent the initiation of the eviction processand, therefore, bar hearings and execution.78 This policy design places the burdenon landlords to comply with the moratoria rather than require tenants toaffirmatively declare that they are covered as is the case under the CDCmoratorium.79 Of course, in rare circumstances landlords may need to pursue an“essential” eviction for conduct that endangers public safety, but evictionmoratorium language should carefully limit this type of exception to ensure thatit avoids all non-essential evictions for non-payment of rent, no cause, and minorlease violations.80

Massachusetts eviction data demonstrates the difference between a policy that

77. See, e.g., ZACH NEUMANN ET AL., EMERGING BEST PRACTICES FOR COVID-19

EMERGENCY RENTAL ASSISTANCE PROGRAMS 11 (2020), https://www.aspeninstitute.org/wp-

content/uploads/2020/12/Aspen-CEDP-Rental-Assistance-Presentation.pdf [https://perma.cc/4PB8-

5MXE] (discussing the effective implementation of rental assistance and the mutually reinforcing

nature of rental assistance and eviction moratoria).

78. See, e.g., Figure 8 (providing data for the number of eviction filings in Massachusetts

during the state moratorium); see also 2020 Mass. Acts. 65.

79. Compare 2020 Mass. Acts 65, with Temporary Halt in Residential Evictions to Prevent

the Further Spread of COVID-19, 85 Fed. Reg. 55,292, 55,292-97 (Sept. 4, 2020).

80. See, e.g., 2020 Mass. Acts 65; see also COLO. REV. STAT. § 13-40-107.5 (2021).


bars the initiation of evictions and one that places the burden on tenants.81 AsFigure 8 illustrates, between April and October 2020, the state barred evictionfilings except in public safety cases, resulting in only sixty-four cases in August2020.82 In mid-October the state’s moratorium lapsed, and landlords could onceagain file eviction cases.83 After the state moratorium lapsed, filing returned topre-pandemic levels.84 Under the CDC moratorium, Massachusetts residents canprevent the execution of an eviction order, but they cannot stop the filing process.Filings dipped again in January and February 2021, perhaps, in part, due tolegislation signed on January 2nd that prevents eviction for anyone awaiting rentalassistance payments from the state agency.85

Figure 8

2. Foreclosure Protections

To ensure the stability of the housing market, foreclosure protections forlandlords can accompany eviction moratoria.86 Because many mom-and-poplandlords will not be able to make mortgage payments or pay property taxes whentenants are unable to pay rent, housing security legislation should mandate thatlenders offer a forbearance option to borrowers unable to make mortgage

81. Mass. Trial Court Dep’t of Research & Planning, supra note 44.

82. Id.

83. Massachusetts Attorney General Issues Advisory to Inform Tenants of Their Rights

During COVID-19 Pandemic, WWLP (Jan. 7, 2021), https://www.wwlp.com/news/massachusetts/

massachusetts-attorney-general-issues-advisory-to-inform-tenants-of-their-rights-during-covid-19-

pandemic/ [https://perma.cc/AR6T-8QDL].

84. Mass. Trial Court Dep’t of Research & Planning, supra note 44.

85. 2020 Mass. Acts 257.

86. See, e.g., Coronavirus Aid, Relief, and Economic Security Act, Pub. L. No. 116-136, §

4022, 134 Stat. 281 (2020).


311

payments due to COVID-19 related hardship, including tenant hardship.Federally backed properties have that option under the CARES Act.87 Foreclosureprotections are critical not only to the stability of mom-and-pop landlords, butalso to ensure that tenants continue living in habitable, safe homes, as landlordsmay not have the motivation to ensure housing quality when homes are inreceivership.

3. Civil Right to Counsel

Similarly, studies have shown a civil right to counsel in eviction cases candeliver significant benefits for tenants, landlords, and society.88 Legal aidattorneys have been proven to reduce disruptive displacement and achieve betteroutcomes for their clients compared to the alternative of lack of representation.In addition to providing representation, lawyers play a crucial role connectingtenants with other social services. At the same time, the success of legal servicesattorneys varies on the features of underlying state law, which sets the terms forcounterclaims and defenses against evictions. While there are no definitive cross-jurisdiction studies that use the same methods to test the underlying law, vastdiscrepancies in recorded outcomes can lead to that hypothesis. In Boston, forexample, tenants with access to full legal representation have a 66% chance ofstaying in their homes.89 In Seattle, the figure is 20%.90 Regardless of the qualityof the underlying law, legal aid attorneys may be increasingly hamstrung by theaccumulation of rental arrears, which deepens the root problem of an evictioncase. In other words, legal assistance is a necessary but not sufficient solution toan eviction case.

4. Civil Debt Conversion

Civil debt conversion is another rarely used but exceedingly powerful legalmeasure to prevent eviction. Civil debt conversion is the transformation of rentaldebt into civil debt, which removes the landlord’s right to evict the tenant on thebasis of that debt. The California legislature passed Assembly Bill 3088, whichprevented landlords from evicting renters for pandemic related rental arrearsduring March to August 2020.91

87. Id.

88. Eviction Right to Counsel Resource Center, STOUT, https://www.stout.com/en/

services/ transformative-change-con su lt in g / ev ic t ion -r igh t-to-counsel-resources

[https://perma.cc/NFY7-4JZQ] (last visited Apr. 8, 2021).

89. James Greiner et al., The Limits of Unbundled Legal Assistance: A Randomized Study in

a Massachusetts District Court and Prospects for the Future, 126 HARV. L. REV. 901, 927 (2013).

90. See TARA COOKSON ET AL., LOSING HOME: THE HUMAN COST OF EVICTION IN SEATTLE

(2018), https://www.kcba.org/Portals/0/pbs/pdf/Losing%20Home%202018.pdf [https://perma.

cc/TBL8-K8NH]

91. 2020 Cal. Legis. Serv. ch. 37 (A.B. 3088) (West).


B. Cash Assistance to Stabilize Tenancies

Cash assistance is the single most important policy lever government can pullto reduce eviction risk. Renters need money to pay their rents. Landlords needmoney to pay their mortgages. And, only rental assistance payments can clear theledger and make both parties whole.

A number of policies designed to get cash into renters’ hands help flatten theCOVID-19 eviction curve, including direct cash transfers like universal basicincome, unemployment insurance, grants to state and local providers, increasesin housing vouchers, and emergency rental assistance can get money to at-riskrenters. Given the accumulation of rental arrears, flexible rental assistance fundsmay offer the greatest hope to clear large arrearages and stabilize futuretenancies.92 The next part of this Article explores the social and public benefitsof rental assistance in depth.

V. EVALUATING THE COSTS AND BENEFITS OF RENTAL ASSISTANCE

Effective policy solutions to the pandemic eviction crisis combine evictionmoratoria with rental assistance.93 Eviction moratoria keep renters in their homesduring the pandemic or other public health crises, and they buy time for theadministration of rental assistance. Rental assistance stabilizes tenancies: it cancure rental arrears and ensure that tenants have the funds to pay rent while theeconomy recovers. It has the additional advantage of stabilizing landlords,making them whole or close to whole and enabling them to service their mortgagedebt and pay property taxes.

Advocates across the country have been calling for both eviction moratoriaand rental assistance, stating that together they stabilize families and landlords.There is significant literature about the benefits of eviction moratoria on keepingrenters in their homes,94 including homelessness prevention and reduction inCOVID-19 spread.95 But to date, there has been only limited analysis of therelative benefits of a sustained federal rental assistance programs, especially onits impact for non-homeless individuals. Elior Cohen, a PhD student at theUniversity of California, Los Angeles, evaluated the effects of rental assistanceamong homeless individuals in Los Angeles County, finding that 80% of housingcosts are offset by benefits within eighteen months for homeless single adults.96

92. See ANDREW AURAND ET AL., NAT’L LOW INCOME HOUS. COAL., NLIHC RESEARCH

NOTE: THE NEED FOR EMERGENCY RENTAL ASSISTANCE DURING THE COVID-19 AND ECONOMIC

CRISIS (2020), https://nlihc.org/sites/default/files/Need-for-Rental-Assistance-During-the-COVID-

19-and-Economic-Crisis.pdf [https://perma.cc/K5A7-U8CB].

93. NEUMANN ET AL., supra note 77.

94. See, e.g., The COVID-19 Eviction Crisis, supra note 34.

95. Id.

96. See, e.g., Elior Cohen, Housing the Homeless: The Effect of Housing Assistance on

Recidivism to Homelessness, Economic, and Social Outcomes 3 (Oct. 19, 2020) (working paper),

https://nlihc.org/resource/housing-assistance-improves-health-and-well-being-reduces-returns-


313

Researchers at the National Low Income Housing Coalition and the Universityof Arizona estimated the homelessness-related costs of eviction associated withCOVID-19, to be between $62 billion and $128 billion, depending on the numberof households estimated to be at risk of eviction.97 The subsequent analysis aimsto fill that gap by providing a directional estimate of the ROI of federal rentalassistance.

In the following sections, this Article lays the foundation for its assessmentof the ROI. It outlines how the COVID-19 Eviction Defense Project (“CEDP”)ROI model works; the six core building blocks of the model; the most sensitiveassumptions in the model; the core scenarios used to estimate rental assistance;and concludes with an analysis of the ROI. The ROI analysis indicates thatfederal rental assistance has a ROI of between 208% to 466%. This partconcludes with the consideration of the costs avoided as a result of thestabilization of these tenancies.

A. CEDP ROI Model Overview

As detailed in Part I, the CEDP model estimates the ROI of rental assistanceby comparing the costs of pandemic-related displacement with quantifiable publicand social costs avoided from that displacement. The model uses data from theend of December 2020. Given the relative lack of financial assistance in theeconomy at this time, the model offers an opportunity to estimate the effects ofrental assistance with fewer confounding variables. Overall, the model producesa directional analysis of the costs of rental assistance compared to the costsavoided by preventing the social costs of eviction and homelessness. Thismethodology can help policymakers understand orders of magnitude: ROIs above200% can be assumed to be significantly better than the alternative. Smalldifferences in ROI calculations should not be heavily weighted in decision-making.

The model considers the costs of up to two years of rental assistance (allrental debt for 2020 and future rent in 2021). Rental payments cover only themonths that a household would otherwise be at risk of eviction. When householdsstabilize economically, the model assumes the government stops paying rentalassistance. Because of the difficulty in estimating the rates of eviction forindividuals awaiting rental assistance, the analysis assumes that all individualswho are projected to receive rental assistance get it instead of being displaced.However, as the Eviction Lab’s statistics prove, implementation and failures aresignificant and lead to thousands of irreversible and non-essential evictions.

The ROI calculation compares these projected costs with the net presentvalue of three years of public and social costs avoided (2021-2023). The three-year timeline was selected based on the best longitudinal data—through the

homelessness [https://perma.cc/FU4C-MD79].

97. DAN THREET ET AL., COSTS OF COVID-19 EVICTIONS 4 (2020), https://nlihc.org/sites/

default/files/costs-of-covid19-evictions.pdf [https://perma.cc/PY76-5BKL].


United States Department of Housing and Urban Development’s (“HUD”) 12-cityFamily Options Survey—on costs of shelter-based care extending three years.98

These costs avoided only relate to people who are projected to recovereconomically or “cure” housing insecurity by finding other affordable housingafter receiving the benefit. The model does not count “costs avoided” for renterswho are not projected to stabilize even though the net effect is a delay in thedemand for social services, which itself may have significant net present value.

The model includes estimates of public and social costs avoided from a rentalassistance program. Public costs include estimates of expected outlays of local,state, and federal funds associated with the outcomes of evictions. Social costsinclude public costs as well as additional costs to individuals, landlords,neighborhoods, and society.

Cost avoided assumptions were only included where there was sufficientevidence to tie the magnitude of costs avoided to homelessness and evictionprevention. This was sufficient to estimate costs avoided in the followingcategories: housing, health care, employment, legal, foster care, landlord costs,foreclosure costs, and coronavirus costs. The costs avoided may be meaningfullylarger than this Article’s model indicates because estimates do not include largecost categories like education (which were especially hard to estimate in apandemic context).

Figure 9

To execute this ROI methodology, this model followed a six-step process,which is outlined in Figure 9.

1. This model included source data for all renters from the AmericanCommunity Survey, HUD’s Comprehensive Housing Affordability

98. See generally DANIEL GUBITS ET AL., U.S. DEP’T. OF HOUS. & URBAN DEV., FAMILY

OPTIONS STUDY: 3-YEAR IMPACTS OF HOUSING AND SERVICES INTERVENTIONS FOR HOMELESS

FAMILIES (2016), https://www.huduser.gov/portal/sites/default/files/pdf/Family-Options-Study-

Full-Report.pdf [https://perma.cc/Q8U5-JSHF].


315

Strategy (“CHAS”) data set, and the Census Bureau’s Week 21Household Pulse Survey.99

2. It estimated the number of renters at risk of eviction accounting foran economic recovery. The number of people at risk of eviction anddisplacement is calculated over time based on the Census Bureau’sHousehold Pulse Survey data on rental arrears and housing insecurity(renters who express slight or no confidence in their ability to payrent next month). The economic recovery estimates are based uponthe Federal Reserve’s projections for the decline in unemploymentrates. Any economic recovery that has already occurred is de factoincorporated into the model because weekly Census Bureau dataaccounts for the recovery. Approximately 25% of renters at risk ofeviction are assumed to stabilize economically between January2021 and the end of 2021.100

3. The model simulated outcomes for renters at risk, estimating howmany renters would face displacement under a rental assistanceregime and how many would be displaced without intervention.101

The model further estimated renter outcomes, including shelter,rehousing, and doubling up for those who were estimated to havebeen displaced. For renters who do not stabilize due to economicrecovery, the model also triangulates to a cure rate of 70% under arental assistance regime, meaning that 30% of eligible renters areassumed not to stabilize even with access to rental assistance andwould be at risk when the funds expire.102 These results werecompared to the results of displacement in the absence of rentalassistance.

4. The fourth building block estimates eligibility for rental assistance,uptake of rental assistance among the eligible population, thepercentage of evictions avoided for people accessing rentalassistance, and the price of rental assistance for this population basedon the pandemic’s estimated impact on their income.103 It assumesrental assistance would cover all 2020 arrears and provide a stipendfor all 2021 rent for renters who are estimated to remain at risk ofeviction in 2021.

5. The model assessed the public and social costs associated with aneviction, drawing on dozens of academic studies about the outcomesof eviction and homelessness and refining estimates of the costs of

99. See infra Part VII.A.

100. See infra Part VII.B.

101. See infra Part VII.A.

102. This means that for those who do not cure due to an economic recovery, 70% will find

an alternate, permanent, and safe housing arrangement if given a year’s worth of rental assistance.

See infra Part VII.C.

103. See id.


these outcomes based on state cost data.104 6. Finally, it compared the two years of rental assistance costs to the

estimated three years of benefits.

B. Scenario Selection

This Article estimates the ROI of three scenarios for federal rental assistancein increasing levels of coverage. Each scenario is based on the framework ofrental assistance provided in the Consolidated Appropriations Act of 2021.105

• Scenario 1 – Federal Emergency Rental Assistance (“ERA”): Thefirst scenario provides a directional estimate of the ROI for theEmergency Rental Assistance authorized by the ConsolidatedAppropriations Act of 2021.106 The appropriation authorized $25billion in rental and utility assistance for households making lessthan 80% of the AMI and aimed to prioritize households that makeless than 50% of AMI or had an unemployment event. Householdseither qualified for unemployment benefits or experienced areduction in income or significant financial hardship due to thecoronavirus. The bill authorized payments for a maximum of twelvemonths with a three-month extension, including back and future rent,and allowed for 10% of funds to be spent on administration.

• Scenario 2 – Expansive Federal Emergency Rental Assistance(“ERA+”): This scenario builds on the ERA scenario. It includes amore expansive qualification criterion, allowing anyone who makesAMI or below (100% of AMI) to qualify. It also includes individualswho are behind on rent but did not lose employment-related incomeduring the pandemic.

• Scenario 3 – Expansive Federal Emergency Rental Assistance +Partial Settlement (“ERA+ & Partial Settlement”): This scenariobuilds on the ERA+ scenario, with one additional criterion: landlordssettle rental arrears at 80% of face value of the rental debt as acondition of receiving funds. This scenario is informed byWashington State’s rental assistance model.107 It also includes anuptake rate of 100% to demonstrate the benefits of complete funding.

C. Findings: Estimated Returns on Investment of Rental Assistance

The model finds a three-year ROI of pandemic emergency rentalassistance of 208% to 466%. This conclusion indicates that paying people’s

104. See infra Part VII.E.

105. H.R. 133, 116th Cong. (2020).

106. Id.

107. See generally WASH. STATE DEP’T OF COMMERCE, GUIDELINES FOR THE EVICTION RENT

ASSISTANCE PROGRAM (ERAP) (2020), https://deptofcommerce.app.box.com/s/kb5sds2gv4

yc9n931j1e7f1dikvlyub1/file/696873212517 [https://perma.cc/Y9X7-RZ46].


317

rents is almost always cheaper than the combined debts and outlays thatmight occur if the government does not foot the bill. More broadly, these ROIestimates indicate that homeless prevention and housing security investments inthe form of rental assistance are good financial investments.

The outlays needed to provide rental assistance at take-up rates of 65% and100% are in the tens of billions of dollars. As Figure 10 illustrates, they rangefrom $56 billion to $92 billion. Governments have struggled to distributefinancial assistance in general108 and rental assistance in particular to individualswho qualify.109 The model leverages the Consolidated Appropriations Act’s 10%allowance for administration to attempt to quantify these costs. Better programdelivery, or payment at full face value, would increase the costs.

As Figure 10 illustrates, the outlays are significant, but the three-year returnson investment are enormous and increasing dependent on relative access. Withoutsuch investments, local, state, and federal governments will have to footenormous social services bills. Public costs associated with homelessness anddisplacement from the pandemic project to devour local, state, and federalbudgets. An investment of $56 billion in a rental assistance program—withsimilar parameters as the Emergency Rental Assistance program in theConsolidated Appropriations Act—would save an estimated $117 billion inquantifiable public costs alone.

108. Ben Zipperer & Elise Gould, Unemployment Filing Failures: New Survey Confirms That

Millions of Jobless Were Unable to File an Unemployment Insurance Claim, ECON. POL’Y INST.

(Apr. 28, 2020), https://www.epi.org/blog/unemployment-filing-failures-new-survey-confirms-that-

m il l ion s -of -job less -we r e -u n a b le - t o - f i l e -an -u n em ploymen t -in su ran ce-c la im /

[https://perma.cc/SGW4-A7HY].

109. Conor Dougherty, Use It or Lose It: Tenant Aid Effort Nears a Federal Cutoff, N.Y.

TIMES (Dec. 15, 2020), https://www.nytimes.com/2020/12/15/business/economy/rental-aid.html

[https://perma.cc/A6CK-K3G3]; see also NEUMANN ET AL., supra note 77.


Figure 10

Scenario 1: Emergency Rental Assistance

ROI 229% - 392%Support: 12.6 million people; 5.3 million households Stabilize: 9.7 million people; 4.1 million householdsEst. Cost: $51 billion

Est. Admin Cost: $5.1 billionEst. Public Cost Savings: $117 billionEst. Social Cost Savings $200 billion

The ROI of the ERA program is estimated at between 208% and 355%.Assuming sufficient funding for all who qualify, this intervention would support12.6 million people in 5.3 million households. Under the economic recovery andcure rate assumptions, the model estimates that this intervention could avoiddisruptive displacement for nearly ten million people in over four millionhouseholds.

This scenario, which estimates the effects of the Emergency RentalAssistance program in the Consolidated Appropriations Act, shows that the mosttargeted rental assistance programs do not stabilize nearly as many renters asmore generous programs. Programs targeted at the lowest income renters canhave higher ROIs due to lower rental outlays on average.


319

Scenario 2: Enhanced Emergency Rental Assistance (ERA+)

ROI - 219-370%Support: 17.7 million people in 7.5 million householdsStabilize: 13.7 million people in 5.8 million householdsEst. Cost: $69 billion

Est. Admin Cost: $6.9 billionEst. Public Cost Savings: $165 billionEst. Social Cost Savings $280 billion

The ROI of an expanded ERA targeted at renters making median income andbelow is between 219% and 370%. This intervention is estimated to support 17.8million people in 7.5 million households. It could stabilize nearly fourteen millionpeople in 5.8 million households This ROI is relatively higher than the firstscenario because federal rental assistance would be accessible to all renters whoneed it and make median income or below regardless of whether they lostsignificant income related to the pandemic. For the group that did not losesignificant income during the pandemic, the costs of providing rental assistanceare modeled at 20% of full rent per month behind, suggesting that rentalassistance is an especially high ROI investment for low-income individuals whoare at the margins between being able to pay rent and not. It also assumes a highertotal outlay of rental assistance driven by broader program applicability.

Scenario 3: 80% Settlements of Rental Debts and Enhanced Emergency RentalAssistance

ROI 275-466%Support: 27.2 million individuals in 11.5 million households Stabilize: 21.1 million individuals in 8.9 million householdsEst. Cost: $89 billion

Est. Admin Cost: $8.9 billion Est. Public Cost Savings: $255 billionEst. Social Cost Savings $433 billion

The model estimates that the ROI of the government paying rental assistancefor all renters experiencing housing insecurity below 100% AMI is between275% and 466%. This intervention is estimated to support 28.7 million people in12.1 million households during the pandemic and stabilize 21.1 million peoplein 8.9 million households into 2022.

This model offers the highest ROI intervention evidencing the additionalvalue that can accrue to governments from offering (or requiring) a settled rateat less than face value. Settlements at less than face value can take rentalassistance dollars further and/or pay for administration.


D. Detail: The Drivers of ROI

The ROI analysis illustrates that it will always be cheaper to pay ahousehold’s rent than to pay for rehousing, health care, foreclosure, landlorddebts, and all of the downstream impacts of displacement and homelessness.

The primary drivers of the ROI analysis are the costs associated withproviding housing and health care to individuals who would otherwise be evictedabsent intervention. As Figure 11 illustrates, housing and health care are thelargest drivers of costs. Combined health care and housing costs account for morethan 80% of the public costs of addressing homelessness and displacement in thismodel.

Among public costs, the largest drivers are the costs of shelter and health carewith foreclosure and job loss costs driving a significant percentage. Shelter andrehousing costs account for approximately 30% of the public costs avoided.Combined hospitalization, physical health care, and mental health care servicesaccount for more than 50% of the costs avoided associated with rental assistancein this model.

The social cost analysis also incorporates losses of wage income, landlordlosses, eviction costs, and broader estimates of the effects of foreclosure onsociety. While the model is not able to capture all of the social costs of eviction,these categories demonstrate the breadth of the impact of eviction on society atlarge.

The cost estimates are primarily driven by the known high costs ofhomelessness. For example, this model was unable to incorporate reliableestimates for the chronic- and mental-health care costs of displacement forindividuals who lose their housing, but who do not become homeless.

Figure 11


321

E. Breakeven Analysis

Given the high proposed ROI, the analysis also considers a break-evenscenario. In other words: even if you believed only these programmaticassumptions, the financial benefits would equal the financial costs. In the case ofpreventing mass displacement, even a program that breaks even has seriousmerits.

The model would reach an ROI of 99.6% under a circumstance where it onlyconsiders the cost avoidance of not paying for shelter or rehousing and hospitalcare for individuals who become homeless or are otherwise displaced as a resultof being evicted. These cost categories were chosen for the breakeven analysisbecause the research and the direct association with eviction is the strongest.110

As long as one considers that that there are additional costs of housingdisplacement beyond the estimated costs of shelter and hospital care andconsiders that peer-reviewed studies underlie these calculations, this analysisshould give comfort that the government will at least break even on its rentalassistance investments.

F. State-By-State

This model is built based on fifty-state data. As a result, the ROIs of rentalassistance vary significantly by state depending on the percentage of renters whoare at risk of eviction in those states and cost variation in certain states. StateROIs are particularly sensitive to average rents and the costs of shelter and healthcare in those states. ROIs range from ~150-400% in Massachusetts andCalifornia, to 380-1128% in Maine. Table 1 includes the state-by-state ROIsusing the public costs of the ERA scenario to set the lower bound and the socialcost of the ERA+ Negotiation scenario to set the higher bound.

110. See infra Part VII.E (describing the data on housing costs and in-patient and emergency

department related costs of homelessness).


Table 1

G. Model Sensitivity

The model is most sensitive to the following five assumptions: 1. The percentage of renter households who recover economically. The

Federal Reserve’s unemployment rate projections in 2020 and 2021inform the modeled economic recovery rate during the rentalassistance period.111

2. The percentage of renter households who “cure” after a year of rentalassistance. The model employs Elior Cohen’s 2020 analysis on theeffect of rental assistance on homeless individuals to estimate thepercentage of renters who do not recover economically will curewith an extra year of rental assistance.112 Cohen found that rental

111. Fed. Open Mkt. Comm., December 16, 2020: FOMC Projections Materials, Accessible

Version, BOARD GOVERNORS FED. RES. SYS. (Dec. 16, 2020), https://www.federalreserve.

gov/monetarypolicy/fomcprojtabl20201216.htm [https:/perma.cc/Z8WW-23B6].

112. See Cohen, supra note 96.


323

assistance was associated with a 20% recidivism rate tohomelessness compared to 40% for those who do not receive rentalassistance.113 The model uses the midpoint between these two pointsand assumes a 30% recidivism rate. To the author’s knowledge, thereis no available data point for the long-term stabilization rate for non-homeless renters who get a year’s worth of rental assistance.

3. The percentage of renter households who would require shelter-based care (and the associated assumption of a right to shelter). Anestimated 21.8% of renters in arrears would require shelter or similarcosts.114

4. The cost of rent by income tier. The model estimates the cost of rentby income tier using custom tabulations from the AmericanCommunity Survey PMS data that estimate the mean rents paid byindividuals based on income.115

5. Shelter costs by geography. Shelter costs are the single mostexpensive cost in this model. They are based on the three-year costsof Usual Care in the Family Options Survey model and scaled bystate based on the average cost of rent in that state.116

H. Implementation

As the take-up rate illustrates, implementation is a tremendous challenge.While a thorough discussion of implementation is beyond the scope of thisArticle, readers should consider a variety of design principles for ensuring hightake-up rates.117 The most successful programs offer easy-to-complete applicationprocesses with low documentation burdens. They are marketed towards low-income and traditionally underserved communities, including communities ofcolor and mom-and-pop landlords who own a disproportionate amount of thenaturally occurring affordable housing. They distribute funds quickly, and theydo so at a negotiated rate rather than full payment—something that would furtherincrease the ROI of rental assistance. The model assumes a 10% cost ofadministration, matching the percentage in the Consolidated Appropriations Act

113. Id.

114. See infra Part VII.C (discussing the estimated distribution of tenants to housing

outcomes).

115. See infra Part VII.B (discussing the methodology for calculating tenant rents).

116. See infra Part VII.E (discussing the methodology for calculating shelter costs).

117. See generally KIM JOHNSON & REBECCA YAE, NAT’L LOW INCOME HOUS. COAL., BEST

PRACTICES FOR STATE AND LOCAL EMERGENCY RENTAL ASSISTANCE PROGRAMS (2021),

https://nlihc.org/sites/default/files/Best-Practices-for-State-and-Local-Emergency-Rental-

Assistance-Programs.pdf [https://perma.cc/8DKW-5EWZ]; NEUMANN ET AL., supra note 77;

Principles for Emergency Rental Assistance During the COVID-19 Crisis, NAT’L HOUS. L. PROJECT

(Jan. 21, 2021), https://www.nhlp.org/wp-content/uploads/ERA-Principles.pdf [https://perma.cc/

JV2N-M67G].


to account for the investment required to create efficient rental assistance deliverychannels.

VI. CONCLUSION

The patchwork of eviction moratoria and rental assistance has largelysucceeded in keeping tenants in their homes, but debts are piling up and evictionsare accelerating. With one in six renters in debt to their landlords, as of the endof 2020, the United States risks an unprecedented displacement and homelessnesscrisis. And, policymakers face a stark choice: pay now or pay much more later.

From a human perspective, we understand that comprehensive policies tokeep tenants in their homes are a no brainer. This Article asked the financialquestion: can we afford such a comprehensive rental assistance program; wouldthe benefits justify the costs?

The answer is an emphatic yes. As this Article illustrates, investing in agenerous rental assistance program now will pay several times over. Andtherefore, nobody in America should face a loss of housing simply because theycannot pay rent for a few months, especially in an economic crisis. The benefitsof a rental assistance program designed to achieve those ends flow to tenants,landlords, and governments alike.

This ROI analysis of pandemic rental assistance presents three coretakeaways for policymakers. First, rental assistance is the rare win-win-win.Tenants get to stay in their homes. Landlords are made whole. And, governmentssave money. Specifically, tenants do not just keep the roof over their heads, theykeep their homes and everything those homes mean to them. The stabilityprevents the serious health and poverty consequences caused by eviction,displacement, and homelessness. Landlords benefit as well. They avoid heavyfinancial losses; mom-and-pop landlords can pay the mortgage and avoid sale orforeclosure. Governments maintain naturally occurring affordable housing andavoid having to make tremendous financial outlays in the form of housing, health,legal, education, and other social costs. For every dollar invested in a rentalassistance program that covers tenants for up to two years, government could see$1.08 in savings (2.08 times the initial investment) and society could see upwardsof $3.66 in benefits (4.66 times the initial investment). The ROI model makes acompelling case that, faced with resource tradeoffs, governments should investheavily in rental assistance.

Second, Congress made a great investment in its Emergency RentalAssistance allocation, and it should invest more and broaden the program. ThisArticle estimates that Congress’ Emergency Rental Assistance program has aquantified ROI of 208-355%. While a good start, more resources and protectionsare needed to stabilize households. To cover the impacted population through theend of 2021, Congress must allocate between $56 billion and $92 billion. Itshould also broaden the program. Broadening the program to serve all who arebehind on rent and make less than median income would increase the estimatedROI to 275-466%.

Third, and more broadly, the ROI analysis makes a compelling case thatnobody in America should face a loss of housing simply because they cannot pay


325

rent. Reasonable people can disagree over what constitutes a “just cause” or“good cause” eviction. However, this analysis emphatically provides that a non-payment eviction should never be deemed “essential.” While our eviction systemhas always been a moral failure—even more so in the midst of a deadlypandemic—the ROI analysis suggests that it is also financial failure. We need asystem focused on stabilizing renters in their homes rather than a system thatsanctions displacement. The benefits would flow to tenants, landlords, andsociety alike. The benefits are financial. More importantly, the benefits arehuman. Money and morality both argue for a reimagination of the social safetynet for renters: with a rental assistance system, policymakers can eliminate non-payment evictions and achieve a rare win for tenants, landlords, and government.

VII. APPENDIX: DRAFT OF METHODOLOGY REPORT FOR RENTAL

ASSISTANCE ROI MODEL

A. Eviction Risk Model Methodology: Overview

The Rental Assistance ROI Model simulates eviction risk, by state and bymonth based on individuals’ self-reported rental debts and their own expressionof their ability to pay rent on time. Estimates were made prior to the passage offederal rental assistance in the Consolidated Appropriations Act and thereforecontemplate eviction risk absent any federal intervention.

At the end of 2020, 18% of renters were already in arrears, and 31% ofrenters had slight or no confidence in their ability to pay rent next month.118 Whena family runs out of money and is unable to be able to pay rent on time, thatfamily is “at risk of eviction.” Of course, not all families who run out of moneywill go through the legal process of eviction or be involuntarily displaced.However, all families that run out of the ability to pay rent, face the risk ofeviction and the likelihood of displacement. My eviction risk estimates includeestimates of involuntary displacement. According to one Matthew Desmond andTracey Schollenberger study of forced displacement in Milwaukee after the GreatRecession, for every formal eviction there are two informal evictions.119

This model is built off of three primary sources. First, its primary input is theWeek 21, Public Use File of the Census Bureau’s Household Pulse Survey.120

Second, it uses the 2019 American Community Survey to establish the numberof renter households nationally, renter incomes, and the mean rent by AMI tier.121

118. See Week 21 Household Pulse Survey: December 9 – December 21, supra note 8.

119. Desmond & Schollenberger, supra note 27.


121. Household Income by Gross Rent as a Percentage of Household Income in the Past 12

Months, U.S. CENSUS BUREAU (2018), https://data.census.gov/cedsci/table?q=ACS%20Table%

20B25074,%201-year%20data%20for%202018&tid=ACSDT1Y2018.B25074&hidePreview=false

[https://perma.cc/4DLC-GYAF]; Public Use Microdata Sample (PUMS), U.S. CENSUS BUREAU,

https://www.census.gov/programs-surveys/acs/microdata.html [https://perma.cc/A9JH-93DN] (last


Third, it uses the Census Bureau’s CHAS data to establish the percentages ofAmericans by AMI tier and cost burden.122

The percentage of renter households at risk of eviction by month is based onthe percentage of renters who indicate housing insecurity.123 The model beginswith the percentage of renters who are housing insecure in the Week 21 PublicUse File for the Household Pulse Survey, which includes data on housinginsecurity and rental debts.124 It assumes that individuals who indicate that theyare already in arrears and housing insecure were already at risk of eviction in2020. Individuals who indicate housing insecurity—but no arrears—are assumedto become at risk of eviction over time in 2021 as their resources run out.Individuals who are in arrears but do not claim housing insecurity are assumedto “cure” and are therefore not at risk of eviction.

The model includes estimates by income tier, finding greater housinginsecurity among lower-income renters. It translates housing insecurity by incomeband into housing insecurity by AMI tier. Because the Household Pulse Surveyreports its responses at the person level for individuals older than eighteen, ratherthan the household level, the model scales responses to the household level withthe assumption that adult (over-eighteen) responses are equivalent to householdlevel responses.125 The model reports these eviction risk numbers in terms ofpeople rather than households. It removes renters who do not report their incomefrom both the numerator and the denominator. It calculates that the average renterfamily size is 2.37 compared to an average family size of 2.53 based on theAmerican Community Survey.126

After estimating the point in time amount of insecurity and eviction risk, themodel estimates when renters fall behind on rent and when they recover. Themodel uses three primary inputs to do this. First, the model assumes a rate ofincrease in the number of people who are behind on rent since March 2020—the

updated Feb. 23, 2021).

122. See generally Consolidated Planning/CHAS Data, OFF. POL’Y DEV. & RES., https://www.

huduser.gov/portal/datasets/cp.html [https://perma.cc/D4T7-V9VA] (last visited Apr. 15, 2021).

123. Housing Insecurity, U.S. CENSUS BUREAU, https://www.census.gov/data-tools/demo/

hhp/#/?measures=HIR [https://perma.cc/FSE9-VQTZ] (last visited Feb. 26, 2021) (defining

housing insecurity as people who answer that they have slight or no confidence that they will be

able to pay rent on time next month or who are not current on rent).


125. Alexander Hermann & Sharon Cornelissen, Housing Perspectives – Using the Census

Bureau’s Household Pulse Survey to Assess the Economic Impacts of COVID-19 on America’s

Households, JOINT CTR. FOR HOUSING STUD. HARV. U. (July 2, 2020), https://www.jchs.harvard.

edu/blog/using-the-census-bureaus-household-pulse-survey-to-assess-the-economic-impacts-of-

covid-19-on-americas-households [https://perma.cc/23BH-PUGQ] (describing the creation of

household weights from person weights and the fact that only a single individual is surveyed per

household).

126. ACS Demographic and Housing Estimates, U.S. CENSUS BUREAU (2018), https://data.

census.gov/cedsci/table?q=United%20States&g=0100000US.04000.001&tid=ACSDP1Y2018.

DP05&vintage=2018&hidePreview=true&moe=false&tp=false [https://perma.cc/7JQ7-RFFN].


327

majority of the increase occurring after the expiration of federal enhancedunemployment benefits in July and August 2020.127 The model employs thePhiladelphia Federal Reserve’s distribution of rental arrears—for those who lostjobs during the pandemic—as source data to estimate the percentage of renterswho fell behind in each month and the percentage of arrears that accrued in thatmonth.128 There is minimal data on the number of renters who fell behind bymonth and the amount of rental debt that has built up in the economy by month,this assumption was chosen as the best among few data points.129

Second, the model establishes an estimated economic recovery as well as therate of increase for eviction risk for housing insecure individuals who do notrecover. The model estimates economic recovery over the next twelve months forapproximately 25% of families at imminent risk of eviction, but families who donot experience an economic recovery ultimately exhaust financial resources topay rent and therefore become housing insecure and at risk of eviction. FederalReserve projections enable the estimate of a monthly “cure” rate of 2.1% permonth in 2021, accounting for the number of people who can stabilize afterreturning to work.130 This economic recovery has the effect of “curing” housinginsecurity in this model, assuming that these families who return to work couldstabilize economically with payment plans. Previous months of economicrecovery are assumed to have been reflected in survey data: families that haverecovered economically—who were previously housing insecure—are presumedto have at least moderate confidence in their ability to pay rent the next month.

Third, for prospective eviction risk, the model selects those who were notbehind on rent on the Household Pulse Survey but indicate slight or noconfidence that they can pay next month’s rent on time, and it assumes that one-sixth of them are at risk of eviction for the first time. This assumption iscountered by the assumption of the economic cure rate. This assumption appliesto both those who lost income and those who did not. By June 2021, allindividuals who are housing insecure are estimated to have fallen behind withoutadditional funds.131

127. Alicia Adamczyk, It’s Been a Month Since the Extra $600 Per Week in Enhanced

Unemployment Benefits Expired. Here’s Where Things Stand Now, CNBC (Aug. 28, 2020),

https://www.cnbc.com/2020/08/28/its-been-a-month-since-enhanced-unemployment-benefits-

ended.html [https://perma.cc/LEA8-CMT0] (discussing the expiration of enhanced unemployment).

128. See generally DAVIN REED & EILEEN DIVRINGI, FED. RESERVE BANK OF PHILA.,

HOUSEHOLD RENTAL DEBT DURING COVID-19 (2020), https://www.philadelphiafed.org/-/media/

frbp/assets/community-development/reports/household-rental-debt-during-covid-19.pdf

[https://perma.cc/3F3V-9GUS].

129. See, e.g., PARROTT & ZANDI, supra note 55 (estimating aggregate rental arrears but not

estimating arrears by month).

130. Fed. Open Mkt. Comm., supra note 111.

131. It is difficult to estimate how long people can continue to pay rent using earned income,

borrowing, and scraping resources from other sources. This estimate assumes that families will

have spent down all additional options outside of additional assistance within six months. See


B. Rental Arrears

1. Context

A variety of actors have attempted to quantify the rental assistance needs in2020 and 2021. Stout produced a variety of measures,132 most recently estimatingthat Americans will have accumulated between $13-24 billion in rental arrears.133

Moody’s Analytics (“Moody’s”) recently updated their estimate of rental debtfrom $70 billion by the end of 2020134 to $57.3 billion.135 Additionally, Moody’sestimates total arrearages of $43 billion by June 2020, inclusive of the federalrental assistance ($68 billion without).136 Both Stout and Moody’s based theirestimates on the Household Pulse Survey and considered rental arrears for “alldelinquent renters” not just those impacted by the pandemic.137 The allocation ofrental arrears by month in each methodology is unclear.

This model estimates that COVID-related rental arrears reached $22 billionin December 2020 and $27 billion in January 2021. These estimates fall at thehigh end of Stout’s range and are approximately half of Moody’ estimates.Discrepancies are likely driven by a difference in the allocation of rental arrearsby month and the fact that Moody’s arrearages also include approximately 24%for utilities and late fees. Accounting for utilities and late fees, CEDP’s estimatesfor January 2021 would reach $33 billion.

By June 2021, this model estimates a rental shortfall of $85 billion and ashortfall of $148 billion by December 2021. The June, $85 billion, shortfall figureis approximately in line with Moody’s estimate ($67 billion accounting for theinclusion of federal stimulus). The primary differential is likely driven bydifferences in predicted economic recovery rates. The rental shortfall figure is anestimate of the amount of rental arrears that could build up in the economy at

MATHIEU DESPARD ET AL., DO EITC RECIPIENTS USE TAX REFUNDS TO GET AHEAD? NEW

EVIDENCE FROM REFUND TO SAVINGS (2015), https://openscholarship.wustl.edu/cgi/viewcontent.

cgi?article=1589&context=csd_research [https://perma.cc/STH6-AVS5] (describing the fact that

only 10% of EITC payments are still available after six months).

132. STOUT, ANALYSIS OF CURRENT AND EXPECTED RENTAL SHORTFALL AND POTENTIAL

EVICTIONS IN THE U.S. 5 (2020), https://www.ncsha.org/wp-content/uploads/Analysis-of-Current-

and-Expected-Rental-Shortfall-and-Potential-Evictions-in-the-US_Stout_FINAL.pdf

[https://perma.cc/7XFT-2MPX].

133. Estimation of Households Experiencing Rental Shortfall and Potentially Facing Eviction,

supra note 52.

134. Mark Zandi, The Week Ahead-U.S. Europe, Asia-Pacific, MOODY’S ANALYTICS (Aug.

13, 2020), https://www.moodysanalytics.com/-/media/article/2021/weekly-market-outlook-market-

value-of-us-common-stock-soars-to-record-high-percent-of-gdp.pdf [https://perma.cc/K8DL-

SLZL].

135. PARROTT & ZANDI, supra note 55, at 4.

136. Id.

137. Id.


329

2021 rates, accounting for our estimated economic recovery. It is important tonote, however, that this figure does not account for evictions—after which debtwould stop accruing—or decreases in rent from 2020 numbers.

2. Methodology

A number of inputs drive these calculations of rental arrears: AMI tier, meanrents by AMI tier, pandemic employment impact category, and estimates of rentalarrears by pandemic employment impact category. For each AMI tier, the modelestimates mean rents and multiplied them by the percentage of individuals in eachpandemic employment impact category (job loss, wage loss but no job loss, andno job or wage loss), and the assumed percentage of rent owed by category

The model establishes average rents by calculating the mean rent paid foreach AMI tier in the American Community Survey.138 Estimates of rents by AMItier are based on the 2019 American Community Survey.139 First, AMIboundaries were established at the 30%, 50%, 80%, 100%, and 150% levels.Average rents by AMI tier were established by calculating average incomes andaverage percentages of income paid to rent by AMI tier. For renter-occupiedhomes where the renter is not paying rent, the “N/A” value was set to zero. Theanalysis considered both mean and median and used mean rents given thelikelihood of skewed, non-normal distribution of rental prices, particularly at thebottom of the market.

Like Stout, the model estimates arrearages, excluding those who live insubsidized housing.140 We estimate that approximately 15% of very low incomehouseholds live in subsidized housing.141 Arrearages for these individuals areassumed to be supported by rental recertification rather than rental assistance.

Next, the model is subdivided based on pandemic employment impactcategory—whether they lost their jobs, income, or did not lose jobs or incomeduring the pandemic using the Household Pulse Survey. This is an imperfectproxy based on the set up of the Household Pulse Survey. The Survey does notask about job loss or wage loss directly, leading to the need to estimate. First,tenants are divided based on to their answer to a question about whether theyhave worked in the last seven days, using the variable “ANYWORK”. For tenants

138. Median HCIR Data by State (unpublished data) (on file with the author) (tabulated by

National Low Income Housing Coalition).

139. See American Community Survey (ACS), U.S. CENSUS BUREAU, https://www.census.

gov/programs-surveys/acs [https://perma.cc/2QR5-79VQ] (last visited Apr. 8, 2021).

140. See Estimation of Households Experiencing Rental Shortfall and Potentially Facing

Eviction, supra note 52.

141. Federal Rental Assistance Fact Sheets, CTR. ON BUDGET & POL’Y PRIORITIES (Dec. 10,

2019), https://www.cbpp.org/research/housing/federal-rental-assistance-fact-sheets#US [https://

perma.cc/X5B4-CPNQ]. This estimate includes the number of households by AMI tier according

to our estimates of AMI tier and the number of households who live in federally subsidized housing

according to the Center on Budget and Policy Priorities.


who answer no they are further subdivided into assumptions of whether tenantslost their jobs or lost income based on the variable “RSNNOWRK.” The modelallocates the following reasons for “job loss”: (1) not wanting to be employed; (2)being sick with COVID-19; (3) caring for someone with COVID-19; (4)concerned about getting/spreading the virus; (5) being retired; (6) getting laid offdue to pandemic, (7) place of employment closed temporarily; and (8) place ofemployment went out of business. Responses about employers experiencing areduction in business, including furlough, due to the pandemic are allocated to the“lost income but not wages” category. The model allocates “other” and “caringfor a child” 50-50% between both categories. The result is: approximately 75%of renters who have lost income are coded as having lost jobs and 25% are codedas having lost income but not their job.

For renters who lost their jobs, they are assumed to owe 100% of rent foreach month in arrears. For renters who lost income but not full employment, theyare assumed to owe 50% of rent for each month in arrears. For renters who didnot lose their job, they are assumed to owe 20% of rent for each month in arrears.

Finally, the model estimates utility arrears using assumptions from theMoody’s rental assistance analysis. It estimates that utility arrears areapproximately 26% of rental arrears by month.142

C. Eligibility, Uptake, Evictions Avoided, and Costs Due to Rental Assistance

1. Context:

A small but growing body of evidence indicates that rental assistance has ahigh ROI and reduces recidivism to homelessness among the populationexperiencing homelessness.143 This evidence, based on individuals who havealready lost their homes, provides a good proxy for the effect of rental assistancein reducing the potential for homelessness among individuals who are housinginsecure but not yet unhoused—the percentage of renter households who “cure”after a year of rental assistance. Elior Cohen’s 2020 analysis on the effect ofrental assistance on homeless individuals in Los Angeles found that rentalassistance was associated with a 20% recidivism rate to homelessness comparedto 40% for those who do not receive rental assistance.144 In other words, 80% oftenants who received rental assistance in Cohen’s study avoided future housinginstability during the study period.

The recently passed Federal Recovery Act helps set parameters for eligibilityassumptions. The appropriation authorized rental and utility assistance forhouseholds making less than 80% of the AMI and aimed to prioritize householdsthat make less than 50% of AMI or had an unemployment event.145 Households

142. PARROTT & ZANDI, supra note 55.

143. The New Leaf Project, FOUND. FOR SOC. CHANGE, https://forsocialchange.org/new-leaf-

project-overview [https://perma.cc/59WK-CY7L] (last visited Feb. 28, 2021).

144. See Cohen, supra note 96.

145. See H.R. REP. NO. 116-68 (2020).


331

either qualified for unemployment benefits or experienced a reduction in incomeor significant financial hardship due to the coronavirus.

There is no good data on rental assistance uptake among those eligible dueto the limited availability of funds and the difficulty of distributing funds duringthe pandemic. States and localities have also created or expanded rental assistanceprograms during the pandemic, leveraging federal and state dollars. Accordingto the National Low income Housing Coalition, states and localities haveallocated $3.9 billion in rental assistance, with at least $2.9 billion coming fromthe CARES Act in 438 different rental assistance programs.146 These programstypically require renters to demonstrate COVID-related economic hardship, proofof residency, income eligibility, among other criteria.147

As a variety of best practice guides reflecting on rental assistance distributionin 2020 have indicated, these documentation burdens can have the effect ofslowing uptake and distribution of strained resources.148 For example, Louisiana’soriginal application required a fifty-five-page application. Despite the burden, thestate closed the application three days after it opened, but disbursementchallenges remained.149 It had disbursed only $115,000 of $5.6 million expectedto go to New Orleans residents. In Massachusetts, the rental assistance programunder the Coronavirus Relief Fund has been described as “strained to the pointof bursting” despite new infusions of state funds.150

The difficulty disbursing rental assistance suggests that it will be difficult toestimate take up rates among eligible renters, and it will be difficult to approachuniversal access for those who are eligible even if sufficient funds exist.

2. Methodology

Rental assistance costs are determined on the basis of eligibility, uptakepercentages, the percentages of evictions avoided due to economic recovery and“cures,” and the formula for calculating rental assistance.

146. See YAE ET AL., supra note 42.

147. Id. at 5.

148. See, e.g., NEUMANN ET AL., supra note 77; JOHNSON & YAE, supra note 117; VINCENT

REINA ET AL., COVID-19 EMERGENCY RENTAL ASSISTANCE: ANALYSIS OF A NATIONAL SURVEY

OF PROGRAMS (2021), https://nlihc.org/sites/default/files/HIP_NLIHC_Furman_Brief_FINAL.pdf

[https://perma.cc/NU7P-HF4N].

149. Michael Isaac Stein, Suspended $24M State Rental Assistance Program Has Only

Disbursed $115,000 in New Orleans, LENS (Sept. 22, 2020), https://thelensnola.org/2020/09/22/

suspended-24m-state-rental-assistance-program-has-only-disbursed-115000-in-new-orleans/

[https://perma.cc/9XKW-EE5N].

150. Tim Logan, Delays and Debt Mount as State Rent Relief Program Strains to Meet

Demand, BOS. GLOBE (Oct. 24, 2020), https://www.bostonglobe.com/2020/10/24/business/delays-

debt-mount-state-rent-relief-program-strains-meet-demand/?p1=Article_Inline_Text_Link

[https://perma.cc/X6CU-UB43].


a. Eligibility

The model considers two types of eligibility criteria: eligibility based on (1)COVID-19 employment impact category and (2) pre-COVID-19 income. Thefirst eligibility scenario assumes that only those who have lost income and lostwages due to COVID are eligible for this rental assistance.151 The secondeligibility assumes that anyone can be eligible. Both scenarios can either assumethat renters of any pre-COVID-19 income level who experienced a COVID-19economic hardship will be eligible for rental assistance or they can be limitedbased on AMI tier.152 As noted in the scenario analysis, and in the abovedescription of federal rental assistance, funds are prioritizing those who make lessthan 80% of AMI.

b. Uptake Percentages

The model estimates that 65% of renters who are eligible will be able toaccess the rental assistance. This estimate is based on an Economic PolicyInstitute study of unemployed workers accessing unemployment and pandemicunemployment assistance.153 The Urban Institute also used this 65% take up rateto estimate take up of rental assistance among those eligible.154 Our model allowsfor estimates of different uptake percentages. Given the documentation burdenof rental assistance programs compared to unemployment filing, 65% is assumedto be the upper bound of uptake and disbursement. Ben Zipperer and Elise Gouldfurther note that the real assistance take up rate may have been between 45-52%.155

c. Evictions Avoided

Evictions Avoided: Evictions avoided are calculated based on eligibility,uptake, economic recovery, and cure rates. For people receiving rental assistance,the model assumes that rental assistance leads to an avoided eviction during themonths that someone is receiving rental assistance. The model assumes that 25%of individuals would permanently avoid eviction based on the economic recoveryin 2021. As outlined above, the model applies an economic recovery rate basedon the Federal Reserve’s projections of improving unemployment rates in2021.156 Additionally, the model estimates that up to twenty-four months of rental

151. See supra Part VII.B (describing how COVID-19 employment impact categories are

established).

152. See supra Part VII.A (describing how the model incorporates AMI tiers).

153. Zipperer & Gould, supra note 108.

154. STROCHAK ET AL., supra note 28, at 5.

155. Zipperer & Gould, supra note 108.

156. See supra Part VII.A (describing the economic recovery assumptions); see also Press

Release, Bd. of Governors of the Fed. Reserve Sys., Chair’s FOMC Press Conference Projections

(Sept. 16, 2020), https://www.federalreserve.gov/monetarypolicy/files/fomcprojtabl20200916.pdf


333

assistance leads 70% of renters to permanently avoid disruptive displacement.157

The model uses Elior Cohen’s 2020 analysis on the effect of rental assistance onhomeless individuals to estimate the percentage of renters who do not recovereconomically but would cure with an extra year of rental assistance. Cohen foundthat rental assistance was associated with a 20% recidivism rate to homelessnesscompared to 40% for those who do not receive rental assistance.158 This analysisuses the midpoint between these two points and assume a 30% recidivism rate.This proxy was used due to the lack of data on the long-term stabilization ratesfor non-homeless renters who get a year’s worth of rental assistance.

d. Costs of Rental Assistance

Costs of Rental Assistance: For each renter, this analysis estimates that rentalassistance pays the tenant’s complete rent. This assumption is based on theConsolidated Appropriations Act, which allows federal funds to cover the fullamount of arrears for each applicant.159 The model calculates rent as a percentageof income as described above.160

D. Potential Outcomes for Those Who Are Housing Insecure

This model allocates all individuals who are housing insecure into estimatedoutcome groups without rental assistance. The allocations follow a decision treethat occurs at three levels: (1) formal v. informal evictions; (2) serial evictions forthose in the formal eviction process; and (3) housing outcomes for those who areformally and informally evicted.

1. Formal v. Informal Evictions

a. Context

This is admittedly a rough assumption based on minimal data. The courts donot and cannot track informal evictions, and there are few studies that measurethese outcomes. Informal evictions are necessarily difficult to measure given thatthey occur outside of the formal legal system.

As noted in the introduction, there are 3.2 million eviction filings in anygiven year.161 Eviction filings are lagging numbers, as many tenants move at the

[https://perma.cc/W8EH-P9YQ].

157. Cohen, supra note 96.

158. See id.

159. H.R. 133, 116th Cong. (2020).

160. See supra Part VII.B (describing the calculation of mean rents by household type).

161. See Map & Data, EVICTION LAB, https://evictionlab.org/map/#/2016?geography=states&

bounds=-159.898,68.074,-72.438,79.561&type=er [https://perma.cc/XDC2-LJYG] (last visited Feb.

28, 2021).


first demand that they pay rent or vacate, or at the first mention of eviction. Forexample, 34% of people who received eviction notices in Milwaukee, movedwithout going to court.162 The same study found that for every one formaleviction in Milwaukee, there were two informal evictions.163

Eviction threats create fear among renters who do not know their rights tolegal process or have other reasons to fear engagement with the legal system.They also lead renters who understand that an eviction can function as a “ScarletE” to vacate their homes prior to a filing, so they can improve their ability to re-rent in the future.164

b. Methodology

Of those at risk of eviction, the model estimates that one third would gothrough the formal process of evictions and two thirds would face informaldisplacement. This assumption draws upon Matthew Desmond and TraceySchollenberger’s 2015 study that uses 2009-2011 data on formal vs. informalevictions in Milwaukee.165 The study finds a 2:1 ratio between informal andformal evictions.166 The model leverages this ratio to estimate that 33% ofhousing insecure individuals may experience a formal eviction filing with thecourts when they exhaust their resources.167 The other 67% are estimated to faceinformal eviction after exhausting their resources.

2. For Formal Evictions: First-Time Displacement, Serial Evictions, andLegal Cures

a. Context:

There is likely to be significant variability in cure rates and rates of serialevictions—rates that are inversely correlated. Where serial eviction filings arehigh, tenants are more likely to cure, only to face eviction filings again in futuremonths. In these states, eviction filings are often disproportionately used as a rentcollection or tenant management tool.168 Using a data set that includes more thaneight million filing records from twenty-eight states, Lilian Leung, PeterHepburn, and Matthew Desmond found significant “within- and between-state

162. @just_shelter, TWITTER (Jan. 28, 2021, 5:51 PM), https://twitter.com/just_shelter/status/

1354925182499102722 [https://perma.cc/8KPV-DRV2].

163. Desmond & Schollenberger, supra note 27, at 1761.

164. The Scarlet E, supra note 68.

165. Desmond & Schollenberger, supra note 27, at 1761.

166. Id.

167. Id.

168. See Lillian Leung et al., Serial Eviction Filing: Civil Courts, Property Management, and

the Threat of Displacement, 99 SOC. FORCES 1 (2020) (describing variability in serial eviction rates

and cure rates).


335

variation” in the rates of serial evictions by jurisdiction.169 They attributed serialeviction filing discrepancies to legal regimes, ownership structures, and propertymanagement strategies.170 Serial evictions were most common in “mid-rangerental markets,” suggesting that even in normal times renters in workforce andmiddle-income communities face eviction risk.171

b. Methodology

The formal eviction estimates are based on three important assumptions. First19% of renters cure a non-payment eviction filing. This assumption comes fromLosing Home: The Human Cost of Eviction in Seattle, finding that 75% of tenantsvacated their unit after a filing.172 To reach this estimate, results where theoutcome was unclear or the tenant died were omitted, leading to the estimate that81% of tenants would vacate after a filing if no rental assistance was provided.The assumption was selected given the clarity of the data and the likelihood thatsignificant rental arrears would ultimately be unrecoverable through an eviction-as-rent-collection strategy. Second, 30.4% of tenants experience a serial evictionfiling—a second filing in our model.173 Third, the model assumes a maximum oftwo eviction filings per year and per tenant household for simplicity. As rentalarrears deepen during the pandemic economic crisis, property recovery maybecome a stronger goal of property owners relative to baseline.

3. For Both: Housing Outcomes Post Eviction

a. Context

A small number of studies have attempted to estimate the outcomes for folkswho are evicted. Robin Hood, for example, estimates that 25% of evicted tenantsin New York end up in shelter,174 a figure cited frequently in Stout’s cost benefitanalysis reports.175 A study of formal evictions in Seattle found that most evictedresidents became homeless with 37.5% unsheltered, 25% in a shelter, 25%doubled up, and 12.5% finding new homes; ultimately nearly half left the city.176

169. Id. at 22.

170. Id. at 2.

171. Id.

172. COOKSON ET AL., supra note 90, at 55.

173. See Leung et al., supra note 168.

174. NEIL STEINKAMP, THE ECONOMIC IMPACT OF AN EVICTION RIGHT TO COUNSEL IN

BALTIMORE CITY 38 (2020), https://cdn2.hubspot.net/hubfs/4408380/PDF/Eviction-Reports-

Articles-Cities-States/baltimore-rtc-report-final-5-8-2020.pdf [https://perma.cc/UG6A-XFJX]

[hereinafter THE ECONOMIC IMPACT]

175. See, e.g., id.

176. COOKSON ET AL., supra note 90, at 60.


b. Methodology

The analysis assumes a right to shelter. The right to shelter is limited in theUnited States. Only a few jurisdictions, including New York City, Massachusetts,and Washington, D.C., have a right to shelter (Massachusetts and Washingtononly have limited rights to shelter for families and in extreme weather conditionsrespectively). And, there are exceptions, like a loss of the right to shelter forresidents of public housing in Massachusetts who have been evicted.177 Althougha tens of thousands of individuals sleep unsheltered every night in the UnitedStates,178 and there is a likelihood of that number increasing withoutinterventions, shelter costs are assumed for each household allocated to the“shelter” category. The social benefit of evictions avoided should not bedepressed on the basis of cities’ failure to have a sufficient number of beds.Moreover, like Stout, estimates of shelter costs can be a helpful proxy for otherrehousing costs necessary to achieve housing stability.179

Potential housing outcomes post eviction consider tenants at risk of bothformal and informal eviction, accounting for cure rates and serial eviction filingas outlined above.

The Cleveland Eviction Study forms the basis of estimates of “what happensnext” to tenants after an eviction filing.180 The study found that 40.5% planned tofind another rental, 16.7% planned to double up, 7.1% said they would livesomewhere else, 9.5% said they would stay at a shelter, and 26.2% did not knowwhere they would stay.181 The model uses these proportions with twomodifications. First, those who say they would live somewhere else are assumedto double up, which has the effect of making cost avoided numbers moreconservative. Additionally, 50% of the individuals who “do not know” areassumed to “double up” and 50% are assumed to end up in shelter.

177. Ruth Bourquin, Basic Shelter Rights (Emergency Assistance), MASSLEGALHELP (Jan.

2011), https://www.masslegalhelp.org/income-benefits/basic-shelter-rights [https://perma.cc/DB43-

N9BA].

178. U.S. DEP’T. OF HOUS. & URBAN DEV., supra note 21.

179. NEIL STEINKAMP, COST-BENEFIT ANALYSIS OF PROVIDING A RIGHT TO COUNSEL TO

TENANTS IN EVICTION PROCEEDINGS 68 (2019), https://info.stout.com/hubfs/PDF/Eviction-Reports-

Articles-Cities-States/Los%20Angeles%20Eviction%20RTC%20Report_12-10-19.pdf

[https://perma.cc/NJK3-QNKG]; NEIL STEINKAMP, ECONOMIC RETURN ON INVESTMENT OF

PROVIDING COUNSEL IN PHILADELPHIA EVICTION CASES FOR LOW-INCOME TENANTS 44 (2018),

h t t ps :/ / c d n 2 . h u bspot . n e t /h u bf s /4 4 0 8 3 8 0 /P D F/C os t -B en ef i t -Im pac t -S tu d ies /

Philadelphia%20Evictions%20Report_11-13-18.pdf [https://perma.cc/NAD9-U9SQ] [hereinafter

ECONOMIC RETURN].

180. APRIL HIRSH URBAN ET AL., CASE W. RES. UNIV., THE CLEVELAND EVICTION STUDY:

OBSERVATIONS IN EVICTION COURT AND THE STORIES OF PEOPLE FACING EVICTION 30 (2019),

https://case.edu/socialwork/povertycenter/sites/case.edu .povertycenter/files/2019-

11/The%20Cleveland%20Eviction%20Study-10242019-fully%20accessible%28r%29.pdf

[https://perma.cc/2NX7-5NRJ].

181. Id.


337

Combining the allocations and accounting for formal, informal eviction, curerates, and serial eviction filings, the model projects that 43.3% of at-risk renterswill find a new apartment, 31.4% will double up, 21.8% will end up in shelter,and 3.5% will permanently cure with their landlord and remain in their unit,without access to rental assistance or economic recovery.

E. Social Costs Avoided for Those Who Access Rental Assistance (Benefits)

After allocating people at risk of eviction into their appropriate outcomegroups, costs were allocated to each group. Recall that tenants were divided basedon whether they were at risk of formal vs. informal evictions, consideringwhether they vacate or do not vacate, and determining the percentage who end upfinding a new home, doubled up with family, or in shelter. For each cost category,the model estimates the net present value of costs avoided and scales all financialfigures to 2019 dollars.

1. Time Period

Like our cost of rental assistance analysis, the net present value of costsavoided (benefits) are considered over a three-year period because the bestavailable data extends for that period of time. The three-year time period ofbenefits assumes that pandemic-related rental assistance is allocated in January2021 and covers a year of future rent and up to a year of rental debts. The three-year estimate is based on the Family Options study, a thirty-seven-monthlongitudinal study estimating the impact of shelter-based care, rapid rehousing,and a number of other housing interventions.182 While eviction moratoria andsome rental assistance likely contributed to costs avoided in 2020, this analysisexcludes those costs.

2. Costs for Families Experiencing Shelter Based Homelessness

Due to the data on the costs of providing services for people experiencinghomelessness, cost estimates for the 21.8% of households estimated to experienceshelter-based homelessness following displacement are much more maturecompared to the remaining 78.2% of households assumed not to experienceshelter-based homelessness. While it is likely that a meaningful percentage ofthose who do not immediately seek help from a shelter may ultimately requirecare from a shelter,183 this possibility was not measured given the difficulty inestimating moves between treatment buckets.

182. See GUBITS ET AL., supra note 98.

183. Molly Scott, Doubled-Up Households: Should They Only Count If They’re "Homeless"?,

URB. INST. (Nov. 18, 2011), https://www.urban.org/urban-wire/doubled-households-should-they-

only-count-if-theyre-homeless [https://perma.cc/DKQ7-STDH].


a. Shelter Based Homelessness Costs

(I) Context

Total household costs and lengths of stays vary widely by location. InPhiladelphia, the total annual cost per individual is $7,340 ($17,394 perhousehold).184 In Baltimore, it is $8,034 per household. In Boston, the householdcost is $26,618;185 and New York’s is $75,949 per year, with the average lengthof stay lasting 425 days.186

(II) Methodology

The model estimates housing costs for individuals experiencing homelessnessby location. It relies on the following inputs

• The number and percentage of households at risk of eviction whowould become homeless if evicted; and

• The average annual cost of shelter by state (scaled by rent).Costs associated with the 21.8% of at-risk households who experience

shelter-based care are calculated based on the Family Options Survey.187 Over athirty-seven-month period, usual emergency shelter care will cost $42,167, a rateof $13,676 per year (excluding inflation).188 Rental costs were scaled by statebased on average rent in that state.

The Family Options Survey was chosen because it provides the mostcomprehensive data set on the costs of shelter-based care across geographies andacross time. The study evaluates four interventions for 2,282 familiesexperiencing homelessness: subsidy-only care, project-based transitional housing,community-based rapid rehousing, and usual care “defined as emergency shelterand housing or services that families can access without an immediate referral toa program that would provide them a place to live.”189 The study evaluatedfamilies in a diverse set of locations: Alameda County, CA; Atlanta, GA;Baltimore, MD; Boston, MA; New Haven, CT; Denver, CO; Honolulu, HI;Kansas City, MO; Louisville, KY; Minneapolis, MN; Phoenix, AZ; and Salt LakeCity, UT.190

184. ECONOMIC RETURN, supra note 179, at 60.

185. Dennis P. Culhane & Thomas Byrne, Ending Family Homelessness in Massachusetts: A

New Approach for the Emergency Assistance Program 4 (May 11, 2010) (unpublished manuscript),

https://repository.upenn.edu/cgi/viewcontent.cgi?article=1152&context=spp_papers

[https://perma.cc/HT2P-NRMW].

186. GISELLE ROUTHIER, COAL. FOR THE HOMELESS, STATE OF THE HOMELESS 2018, at 10, 26

(2018), https://www.coalitionforthehomeless.org/wp-content/uploads/2018/03/CFHStateofthe

Homeless2018.pdf [https://perma.cc/9XUF-ST7X].

187. See GUBITS ET AL., supra note 98.

188. See id.

189. Id. at xvi.

190. Id. at iii.


339

b. Inpatient Medical Costs

(I) Context

Inpatient Medical Costs for individuals experiencing homelessness who getshelter-based care also vary by location. Stout has estimated Inpatient Medicalcosts in Philadelphia and Baltimore among other cities, finding a wide range incosts. In Philadelphia, the cost per visit was estimated at $9,000;191 in Baltimore,it was nearly $38,000.192 In Denver, based on our estimates, we find that inpatientcare costs approximately $16,000 per visit.

(II) Methodology

The model estimates inpatient medical costs for individuals experiencinghomelessness. It relies on several inputs.

• The number and percentage of households at risk of eviction whowould become homeless if evicted;

• The percentage who would require inpatient services in a year andthe average number of visits;

• The percentage of visits attributable to the experience ofhomelessness; and

• The net present value of the cost per visit by state.The methodology used in the CEDP risk model parallels Stout’s various

access to counsel cost-benefit analysis reports to estimate the costs associatedwith inpatient medical care. The model assumes 23% of individuals in householdsexperiencing homelessness will require an inpatient medical visit193 of an averageof seven days.194 Further, 80% of these visits are attributable to the experience ofhomelessness.195 Daily costs of an inpatient visit come from the Kaiser FamilyFoundation.196 Multiplying these figures together yields an average cost perhousehold per year of $7,500 per year.

191. ECONOMIC RETURN, supra note 179, at 46-47.

192. THE ECONOMIC IMPACT, supra note 174, at 44.

193. See generally Margot Kushel et al., Factors Associated with the Health Care Utilization

of Homeless Persons, 285 JAMA 200 (2001).

194. See generally Monica Bharel et al., Health Care Utilization Patterns of Homeless

Individuals in Boston: Preparing for Medicaid Expansion Under the Affordable Care Act, 103 AM.

J. PUB. HEALTH S311 (2013).

195. Kushel et al., supra note 193, at 200.

196. Hospital Adjusted Expenses Per Inpatient Day, KAISER FAM. FOUND. (2018), https://

www.kff.org/health-costs/state-indicator/expenses-per-inpatient-day/?currentTimeframe=

0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22:%22asc%22%7D

[https://perma.cc/2PZK-NNZM].


c. Emergency Room Costs (applied to low- and high-cost scenarios)

(I) Context

Emergency Room Medical Costs for individuals experiencing homelessnesswho get shelter-based care also vary by location. They are significant as unhousedindividuals do not typically have access to other forms of care. In Baltimore,Stout estimated these costs at $7,602 per visit.197 In Philadelphia, the costs werelower at $920 per visit.198

(II) Methodology

The model estimates the cost of emergency room visits associated withindividuals experiencing homelessness as a result of eviction. The model relieson several inputs:


• The percentage who would visit the emergency room in a year andthe average number of visits;

• The percentage of visits attributable to the experience ofhomelessness; and

• The net present value of the cost per visit by state.The emergency room cost methodology is similar to the in-patient care

methodology and largely tracks Stout’s methodology.199 It draws on MargotKushel et al.’s findings that 32% of homeless individuals visit emergency roomseach year,200 for an average of four days per year.201 There are 2.37 people in theaverage renter household nationwide. And an estimated 75% of these visits wereas a result of the experience of homelessness.202 To establish costs, these figuresare multiplied by fifty-state estimates of the costs of an emergency room visit.203

197. See generally THE ECONOMIC IMPACT, supra note 174.

198. ECONOMIC RETURN, supra note 179, at 60.

199. See, e.g., id.

200. Kushel et al., supra note 193.

201. Bharel et al., supra note 194.

202. See generally Margot Kushel et al., Emergency Department Use Among the Homeless

and Marginally Housed: Results from a Community-Based Study, 92 AM. J. PUB. HEALTH 778

(2002).

203. Alia Paavola, The Most, Least Expensive States for an ER Visit, BECKER’S HOSP. REV.

(Dec. 7, 2020), https://www.beckershospitalreview.com/rankings-and-ratings/the-most-least-

expensive-states-for-an-er-visit.html [https://perma.cc/VJ8D-FWRK].


341

d. Foster Care Costs (applied in high-cost scenarios)

(I) Context

Homelessness is associated with greater risk of involvement with the child-welfare system compared to other low-income parents.204 For example, ananalysis of longitudinal data in the National Survey of Child and AdolescentWell-Being found that “inadequate housing” contributed to out-of-homeplacement in 16% of intact families under investigation.205 In other words,housing is a significant contributor to foster care placement even when a familyunit is still together.

(II) Methodology

The model estimates the percentage of children who would require foster carebased on living in households who would be at risk of homelessness that wouldrequire shelter-based services if they were evicted.

• The number and percentage of households at risk of eviction whowould become homeless if evicted and have children, including thenumber of children;

• The percentage who would be put in foster care; and• The net present value of the cost per year.Among the 21.8% of at-risk households who may be at risk of shelter-based

care, 46% of housing insecure households have children.206 Among renterhouseholds with children, the average number of children is 2.1.207 Nearly 4% ofchildren from evicted families are placed in foster care.208 This may be anunderestimate of costs, as the analysis only considers costs for childrenexperiencing homelessness and does not consider other forms of inadequatehousing. Haksoon Ahn et al. provide estimates of minimum adequate foster carecosts for children in the United States: they include $859 per month for childrenbetween the ages of four and thirteen ($749 for children zero through four and

204. Jennifer Culhane et al., Prevalence of Child Welfare Services Involvement Among

Homeless and Low-Income Mothers: A Five-Year Birth Cohort Study, 30 J. SOC. & SOC. WELFARE

79, 79 (2003).

205. Patrick J. Fowler et al., Inadequate Housing Among Families Under Investigation for

Child Abuse and Neglect: Prevalence from a National Probability Sample, 52 AM. J. COMMUNITY

PSYCHOL. 106, 106 (2013).

206. Week 20 Household Pulse Survey: November 25 – December 7, U.S. CENSUS BUREAU

(Dec. 16, 2020), https://www.census.gov/data/tables/2020/demo/hhp/hhp20.html [https://perma.cc/

99XJ-SDTZ].

207. ACS Demographic and Housing Estimates, U.S. CENSUS BUREAU (2018), https://data.

census.gov/cedsci/table?g=0100000US&tid=ACSDP1Y2019.DP05 [https://perma.cc/EZ7D-3VL3].

208. Lisa Berg & Lars Brannstrom. Evicted Children and Subsequent Placement in Out-of-

Home Care: A Cohort Study, 13 PLOS ONE art. e0195295, at 6 (2018).


$941 for children fourteen to eighteen).209 The multiplication translates into$10,188 per year per child before inflation.

e. Mental Health Care Costs

(I) Context

Evictions are known to cause increases in mental health issues.210 A 2017meta-analysis of studies on eviction and health concluded that “general consensusthat individuals under threat of eviction present negative health outcomes, bothmental (e.g. depression, anxiety, psychological distress, and suicides) andphysical (poor self-reported health, high blood pressure and child maltreatment),”finding negative mental health outcomes in eleven cohort studies, eight individualcross-sectional studies, and three ecological studies.211

(II) Methodology

The model estimates the financial impact of homelessness-related mentalhealth issues. It uses the following inputs.


• The percentage of chronic health problems associated withhomelessness;

• The number of quality adjusted life years from avoiding chronicmental health problems; and

• The value of a quality adjusted life year.The model draws upon Robin Hood’s methodology to calculate the social

cost of eviction-related mental health issues.212 Robin Hood estimates a twenty-nine-percentage point differential in the likelihood of experiencing depressionamong homeless children compared to low-income children, based on researchconducted by the National Center for Family Homelessness.213 They, Stout, andthis model use the Robin Hood rate for both homeless children and adults.214 The

209. See Haksoon Ahn et al., Estimating Minimum Adequate Foster Care Costs for Children

in the United States, 84 CHILD. & YOUTH SERV. REV. 55 (2018).

210. See, e.g., Rilwan Babajide et al., The Effects of Eviction on People in Middlesex County,

PARTNERSHIP FOR STRONG COMMUNITIES (June 28, 2016), https://www.pschousing.org/news/

effects-eviction-people-middlesex-county#:~:text=Additionally%2C%20most%20individuals%

20lose%20feelings,It's%20a%20feeling%20of%20helplessness [https://perma.cc/3B7E-AR86].

211. Hugo Vasquez-Vera et al., The Threat of Home Eviction and Its Effects on Health

Through the Equity Lens: A Systematic Review, 175 SOC. SCI. & MED. 199, 202 (2017).

212. See generally ROBIN HOOD, METRICS EQUATIONS (2014), https://robinhoodorg-

production.s3.amazonaws.com/uploads/2017/04/Metrics-Equations-for-Website_Sept-2014.pdf

[https://perma.cc/DV3Y-QLAM].

213. Id. at 82.

214. Id.; ECONOMIC RETURN, supra note 179.


343

estimated value of avoiding mental illness is 0.33 of a quality-adjusted life year(“QALY”), and a QALY has an estimated value of $50,000. Applying thiscalculation to each household member, the total annual mental health cost perhousehold, adjusted for inflation, is $12,521.

f. Chronic Health Care Costs

(I) Context

Evictions and homelessness have also been associated with chronic physicalhealth issues.215 Homelessness and eviction are associated with chronic illnessbased on the same National Center for Family Homelessness study cited by RobinHood.216 A 2016 metanalysis found sixteen studies linking eviction and the threatof household instability to negative health outcomes.217

(II) Methodology

The model estimates the cost of chronic health care for individuals who areevicted and as a result experience homelessness. The model employs severalinputs:


• The percentage of chronic health problems associated withhomelessness;

• The number of quality adjusted life years from avoiding chronichealth problems; and

• The cost of a quality adjusted life year.These estimates draw upon Robin Hood’s methodology to estimate the social

cost of eviction-related chronic illness. Robin Hood estimates a 0.1 QALY valuefor reduction in chronic illness on the basis of a QALY difference between thosewith well controlled and non-well-controlled asthma.218 They indicate that this islikely “a conservative guess” given the potential for more than one chronicillness.219 They extrapolate this rate to apply to the whole population.220 Theestimated 2014 price of a QALY is $50,000. This model estimates a seven-percentage point decrease in chronic illness based on the same National Centerfor Family Homelessness study.221 Combined, the total annual costs of chronicphysical health care associated with an eviction, inflation adjusted, are estimated

215. Vasquez-Vera et al., supra note 211, at 202.

216. Id.

217. Id.

218. ROBIN HOOD, supra note 212, at 81-82.

219. Id.

220. Id.

221. Id.


to be $915.84 per household.

g. Job Loss

(I) Context:

Typically, people understand job loss to be a cause of evictions. In 2016,Matthew Desmond and Carl Gershenson released an analysis that found that jobloss is not just a cause of evictions but also a consequence of evictions.222

(II) Methodology

The model estimates the social and public costs of job loss related toevictions for individuals who did not lose their jobs during COVID. The modelis based on five inputs:

• Population of individuals by state who are at risk of eviction andwould face non-shelter-based displacement;

• The percentage of individuals who still employed among those atrisk of eviction;

• The percentage decrease in employment associated with an eviction;• The number of months unemployed; and• Income replacement by AMI tier.The population of individuals by state who are at risk of eviction and the

percentage of individuals still employed are estimated using the CEDP EvictionRisk Model.223 In other words, this estimate only considers people who are at riskof eviction and whose eviction could trigger job loss.

The percentage decrease in employment associated with an eviction isestimated to be 17%. This figure is the midpoint of the estimated effect ofeviction on employment identified in Matthew Desmond and Carl Gershenson’s2016 study topic.224 This is likely a conservative number on two dimensions: itonly uses the midpoint Desmond-Gershenson range, and the analysis does notapply to those who lost wages but not their jobs.

Unemployment following job loss is assumed to be six months based onJanuary 2021 data from the Federal Reserve Bank of St. Louis.225

To calculate social costs, the model multiplies the number of unemployedrenters who faced eviction and lost their job as a result by the weighted averagemonthly income for each AMI tier. To calculate public costs, it considers the

222. Desmond & Gershenson, supra note 63, at 67.

223. See Desmond & Gershenson, supra note 63, at 61-64 (discussing the methodology for

calculating the percentage of individuals at risk of eviction and for a description of pandemic

employment impact categories).

224. Id. at 60.

225. Average Weeks Unemployed, FED. RES. ECON. DATA, https://fred.stlouisfed.org/series/

UEMPMEAN [https://perma.cc/CLB4-TTNW] (last visited Feb. 27, 2021) (detailing twenty-six

weeks as the average number of weeks unemployed in January 2021).


345

unemployment outlays by state and the lost income tax to the state. It estimatesthat the public cost of unemployment is approximately 70% of prior wages. Stateunemployment replaces approximately 50% of wages,226 and the state and federalgovernment lose approximately 19% of wages in the form of tax income.227

h. Criminal Costs

(I) Context

The experience of homelessness has also been associated with increasedinvolvement with the criminal system.228 It is critical not to draw a causalassociation between homelessness and criminal involvement. Indeed, theexperience of homelessness is associated with police interaction, includinginstances where police officers enforce statutes and ordinances designed toprevent individuals experiencing homelessness from resting in a given location.229

Nevertheless, and often as a result of this policing, criminal system costs aredisproportionately higher for populations experiencing homelessness than otherlow-income populations. Cohen’s study, for example, found that formerlyunhoused individuals receiving housing assistance were associated with a 7.9percentage point lower probability of being charged with a crime and a thirteenday decrease in the average number of jail days compared to unhousedindividuals who did not receive housing assistance.230 It also found a non-statistically significant association with reduced probation.231

226. Policy Basics: Unemployment Insurance, CTR. ON BUDGET & POL’Y PRIORITIES,

h t tps:/ /www.cbpp.org/research/economy/policy-basics-unemployment-insurance

[https://perma.cc/NN3J-F68E] (last updated Jan. 4, 2021).

227. Adam McCann, 2020’s Tax Burden By State, WALLETHUB (June 24, 2020),

https://wallethub.com/edu/states-with-highest-lowest-tax-burden/20494 [https://perma.cc/W3W6-

Z7FL].

228. See, e.g., BROOKE SPELLMAN ET AL., U.S. DEP’T OF HOUS. & URBAN DEV., COSTS

ASSOCIATED WITH FIRST-TIME HOMELESSNESS FOR FAMILIES AND INDIVIDUALS 29 (2010),

https://www.huduser.gov/portal//publications/pdf/Costs_Homeless.pdf [https://perma.cc/VJ9Z-

QQT7]; Dennis Culhane et al., Public Service Reductions Associated with Placement of Homeless

Persons with Severe Mental Illness in Supportive Housing, 13 HOUS. POL’Y DEBATES 107, 137

(2002).

229. See generally TONY ROBINSON & ALLISON SICKELS, NO RIGHT TO REST: CRIMINALIZING

HOMELESSNESS IN COLORADO (2015), https://denverhomelessoutloud.files.wordpress.com/2016/03/

no-right-2-rest.pdf [https://perma.cc/Y27U-F3YB].

230. Cohen, supra note 96, at 106.

231. Id.


(II) Methodology

The model estimates criminal system costs associated with the policing ofhomelessness. The model relies on the following inputs:

• The percentage of adults at risk of eviction who would requireshelter-based care;

• The percentage estimated to become involved with the criminalsystem; and

• The typical cost of criminal system involvement for adults.The model begins by estimating that 36% of adults of households who

require shelter services as a result of a displacement would become involved withthe criminal system.232 The average number of adults per renter household is 1.54.This estimate leverages Elior Cohen’s experimental data on rental assistance andcriminal system costs in Los Angeles to estimate that the eighteen-month costsavings of criminal system involvement that could be avoided through effectiverental assistance is $1,724 and therefore the twelve-month costs avoided are$1,149.233

i. Landlord Eviction and Displacement Costs

(I) Context

While the majority of this piece has focused on the costs of evictions totenants and government, it is important to highlight the cost of non-payment ofrent to landlords. When landlords move to evict or displace tenants, they often doso by accepting complete losses on the property and the costs of turning over theproperty. The National Apartment Association estimates that the typicalcollections rate may be less than 10% for affordable housing.234

(II) Methodology

The model offers conservative cost estimates of the cost to landlords ofevicting tenants. These costs are only included in estimates of the “social cost”of eviction. The model includes the following inputs.

• Estimated uncollected rental arrears;• Costs of changing the locks as a low-range proxy for property

damage;• Turnover costs estimated as one month of rent; and• Court costs and legal fees borne by the landlord for the

approximately one third of tenants who this model estimates wouldbe face eviction through the legal system.

The biggest driver of these cost estimates is “uncollected rental arrears.” The

232. ROBINSON & SICKELS, supra note 229.

233. Cohen, supra note 96, at 38.

234. Multifamily Debt Collections: Best Practices for Owners and Managers, supra note 11.


347

analysis pulls upon the rental shortfall analysis that estimates the total potentialrental arrears in the economy.235 Based on a National Apartment Associationreport, the model assumes that landlords will collect only 10% of rental arrears;90% of rental debts prior to a displacement are counted as losses.236 This numbermay be conservative absent federal rental assistance given the massive arrearagesduring the pandemic.

The model assumes for simplicity purposes that all at risk renters would bedisplaced by the end of 2021 assuming a constant rate of displacement. Itconstructs the counterfactual by assuming that all renters at risk of eviction wouldbe displaced in 2021 absent rental assistance. It assumes that displacement wouldoccur at an even rate, with 8% of potential displacements occurring over eachmonth. This assumption simplifies because it is impossible to estimate howdisplacement would proceed without eviction moratoria in place given the paucityof data on displacements during the pandemic and the unprecedented number ofevictions that could occur. This assumption limits the landlord costs only to rentaldebt and not rental shortfall.

Property costs are conservatively estimated at the cost of changing thelocks—though, landlords are likely to face higher costs due to wear and tear,property damage, and renovation costs. The model does not include moving costs.

Vacancy costs are set at one month’s rent. Apartment vacancies have beenrising in the pandemic,237 and rents have declined overall.238 However, a closerlook at the data shows sustained increases in rental prices for “low” and“moderate” quality homes.239 Homes at the bottom-end of the rental market arecontinuing to see rental price increases,240 demonstrating high demand and a lackof supply of affordable housing.241

Estimates of legal fees and court costs vary meaningfully by jurisdiction. Forthe purposes of this model, they are derived from a triangulation between the2018 Transunion analysis of the costs of eviction242 and the American Tenant

235. See supra Part VII.B (describing methodology for estimating rental arrears).

236. Multifamily Debt Collections: Best Practices for Owners and Managers, supra note 11.

237. JOINT CTR. FOR HOUS. STUDIES OF HARVARD UNIV., THE STATE OF THE NATION’S

HOUSING 2020, at 32 (2020), https://www.jchs.harvard.edu/sites/default/files/reports/files/Harvard_

JCHS_The_State_of_the_Nations_Housing_2020_Report_Revised_120720.pdf [https://perma.cc/

HA75-FSJP].

238. Id.

239. Id.

240. Id.

241. See NAT’L LOW INCOME HOUS. COAL., THE GAP: A SHORTAGE OF AFFORDABLE HOMES

4 (2020), https://reports.nlihc.org/sites/default/files/gap/Gap-Report_2020.pdf [https://perma.cc/

47J2-BCVY] (discussing the fact that there are only 36 units available for every 100 tenants in need

of affordable housing).

242. Andrea Collatz, The True Cost of an Eviction, TRANSUNION (Nov. 2, 2018), https://www.

mysmartmove.com/SmartMove/blog/true-cost-eviction.page [https://perma.cc/7M29-BGFY].


Screen’s analysis.243

2. Non-Shelter-Based Displacement Costs

There is considerably less data available about the social costs and relativeeffects of increased housing insecurity not associated with homelessness, so ouranalysis of non-shelter-based homeless costs is much more limited. Thus, themodel considers only three sets of costs associated with displaced individualswho do not require shelter-based homelessness: (1) the cost of finding a newhome for those allocated to that outcome; (2) increased health care costs fromhaving non-affordable housing; and (3) job losses as a result of the displacementfor those who remain employed (and did not lose wages during the pandemic).Of course, eviction and the threat of eviction have been linked to other negativeconsequences, including chronic physical and mental health care.244 While thisanalysis is unable to estimate those costs, it bears mention that those costsexist—and their existence likely increases the ROI of stabilizing households.

a. Costs of Finding a New Home

(I) Context

State laws and conventions for the cost of rehousing vary widely.245 Inpractice, tenants are often required to pay at least one month’s rent in addition toanother month’s rent for a security deposit. Other geographies require at least twomonths’ rent and/or the payment of first month’s rent, last month’s rent, anddeposit.

(II) Methodology

The model assumes that each renter household, on average, that must find anew home will have to pay first month’s rent and a security deposit equivalent toone month’s rent costs. While some localities and rental properties require threemonths’ (first, last, and security deposit), this model employs the conservativeassumption of only two months’ worth of funds to move into a new apartment.While most assumptions in this model operate at the person level, this operatesat the household level. The estimates for the costs of finding a new home arebased on three assumptions.

243. How Much Will It Cost to Evict My Tenants?, AM. TENANT SCREEN, https://

americantenantscreen.com/how-much-will-it-cost-to-evict-my-tenants/ [https://perma.cc/A827-

2PT5] (last visited Feb. 28, 2021).

244. Vasquez-Vera et al., supra note 211, at 202.

245. Security Deposit Limits, State-by-State, NOLO, https://www.nolo.com/legal-

encyclopedia/chart-security-deposit-limits-state-29020.html [https://perma.cc/H2XU-VHUC] (last

updated Dec. 9, 2021).


349

• The number of households at risk of eviction estimated to avoidshelter-based homelessness;

• The average cost of rent by state; and • The number of households at risk are based on the CEDP Eviction

Risk Model.246 The average cost of rent by state is based on theAmerican Community Survey’s rental data by state.247

b. Health Care Costs

(I) Context

Lack of affordable housing has been linked with poor health and limitedability to seek health care.248 Substandard housing, including residential crowdingthat might be caused by doubling up, has been linked to health issues.249 Lowincome families struggling to pay rent, for example, are less likely to regularlyaccess the same medical care.250

(II) Methodology

The model estimates health care costs for those who experience non-shelter-based displacement. It relies on two primary inputs:

• The number of individuals in households at risk of eviction whowould not experience shelter-based displacement; and

• The cost of health care for those who do not live in affordablehousing.

These estimates are based on a study of the effects of affordable housing onhealth care usage in Oregon.251 In total, various affordable housing interventionsdecreased Medicaid expenditures by 12%, while the usage of outpatient careincreased and emergency department care decreased.252 While the studyinvestigated a number of types of affordable housing interventions, including

246. See supra Part VII.A.

247. See supra Part VII.B.

248. Lauren Taylor, Housing and Health: An Overview of the Literature, HEALTH AFFAIRS

(June 7, 2018), https://www.healthaffairs.org/do/10.1377/hpb20180313.396577/full/ [https://perma.

cc/SV4H-7A69].

249. See Claudia D. Solari & Robert D. Mare, Housing Crowding Effects on Children’s

Wellbeing, 41 SOC. SCI. RES. 464 (2012).

250. Paula Braveman et al., Housing and Health, ROBERT WOOD JOHNSON FOUND. (May

2011), https://www.rwjf.org/content/dam/farm/reports/issue_briefs/2011/rwjf70451 [https://perma.

cc/T78P-BWVD].

251. See generally BILL WRIGHT ET AL., HEALTH IN HOUSING: EXPLORING THE INTERSECTION

BETWEEN HOUSING AND HEALTHCARE (2016), https://www.enterprisecommunity.org/download?

fid=5703&nid=4247 [https://perma.cc/SXV5-LVNA].

252. Id.


permanent supportive housing and housing for seniors with disabilities, thisanalysis only considers family housing—which is most analogous to the pre-COVID-19 housing context of the majority of the demographic. The study founda comparably smaller effect on expenditures. When selecting the family housingoption, the analysis finds a $22 monthly decrease in the amount of Medicaidspending per person.253 The study found a strong association, but it is notstatistically significant at the 5% or 10% level (p value of 0.12%); nevertheless,this study offers a reasonable estimate of the comparative health benefits ofaffordable housing compared to necessarily less affordable housing absent rentalassistance.

c. Job Loss Costs

Both the context and the methodology are the same for cost estimates of jobloss costs for individuals who are displaced but do not experience shelter-basedhomelessness, as they were under the shelter-based homelessness category.

d. Landlord Eviction and Displacement Costs

Both the context and the methodology are the same for landlord eviction anddisplacement cost estimates for individuals who are displaced but do notexperience shelter-based homelessness, as they were under the shelter-basedhomelessness category.

3. Broader Population Costs

The model also considers two cost categories that impact broader society:foreclosure and COVID-19 hospitalizations.

a. Foreclosure Costs

(I) Context

Evictions do not occur in a vacuum. When tenants cannot pay their rent,landlords—particularly mom-and-pop landlords—cannot pay their mortgage.Failure to pay mortgages leads to foreclosure and the accompanying financial andsocial consequences for the community. While likely understated, this modelincorporates an analysis of foreclosure-related risk to demonstrate that failure toprovide rental assistance leads to cascading consequences in the housing marketin addition to the human suffering for evictees, as well as massive financialoutlays for state and local governments.

253. Id.


351

(II) Methodology

The model also includes an analysis of potential foreclosure costs for familiesliving in one- to four-unit rental properties, with mortgages that are owned byindividual investors. This approach broadly follows the methodology charted bythe Harvard Joint Center for Housing Studies.254 The Center’s analysis foundsignificant risk in small buildings and a higher concentration of renters at risk ofCOVID-19 related job loss, as these buildings offer a high percentage of naturallyoccurring affordable housing.255 This analysis is likely to be a conservativepresentation of the potential foreclosure-related costs because a rent-relatedforeclosure crisis would likely affect properties larger than one- to four-units aswell. The analysis finds 5% of landlords at risk of an eviction-related foreclosure.

The public and social costs of eviction-related foreclosure by state are basedon two inputs:

• The number of properties owned by at-risk mom-and-pop landlordsmeasured by (a) the percentage of individual investors who own oneto four properties (b) that rent to non-paying renters and (c) have amortgage (d) where the debt service coverage ratio is less than 80%,assuming (e) half are at risk;256 and

• The Public and Social Cost of Foreclosure measured by the netpresent value of the cost of a foreclosure, estimated from GreatRecession studies of the foreclosure crisis.

To calculate foreclosure costs, the model considers the addressablepopulation257 who live in one to four family units owned by individual investors,selecting the population with a mortgage and a debt service coverage ratio under80%. All data comes from the Rental Housing Finance Survey.258 Theseassumptions culminate in a calculation that 5% of displaced renters live in a unitthat could be at risk of foreclosure. While likely an understatement, this analysisis illustrative of the systemic risk posed by a rental eviction crisis.

The costs of foreclosure were well-studied during and after the last housingcrisis. The Senate’s Joint Economic Committee estimated the social cost offoreclosure to be $77,935, based on costs to financial institutions, cities andlocalities, and the homeowners.259 In 2019 dollars, that total cost is $96,095. Totalsocial costs also included costs to homeowners, renters, state and localgovernments, as well as neighborhood cost. Total public costs were estimated at

254. Airgood-Obrycki & Hermann, supra note 50.

255. Id.

256. Rental Housing Finance Survey (RHFS), U.S. CENSUS BUREAU, https://www.census.

gov/programs-surveys/rhfs.html [https://perma.cc/3UN7-C6LV] (last visited Feb. 28, 2021).

257. Week 20 Household Pulse Survey: November 25 – December 7, supra note 206.

258. Rental Housing Finance Survey, supra note 256.

259. CHARLES E. SCHUMER, CHAIRMAN, JOINT ECON. COMM., SHELTERING NEIGHBORHOODS

FROM THE SUBPRIME FORECLOSURE STORM 16 (2007), https://www.jec.senate.gov/archive/

Documents/Reports/subprime11apr2007revised.pdf [https://perma.cc/QVK3-T97S].


$23,537 per home on the basis of the costs to state and local government offoreclosure and excluding costs to financial institutions and the homeownersthemselves.

b. COVID-19 Costs

(I) Context

A growing body of research has linked evictions to the spread of COVID-19and therefore to COVID-19 related fatalities. In early 2021, researchers at theNational Bureau of Economic Research found, for example, that comprehensivebans on evictions could have reduced infections by 14.2% and deaths by 40.7%during the crisis.260 Researchers at the University of Pennsylvania261 and teamscollaborating with Emily Benfer,262 have also linked eviction to COVID-19infections and deaths.

(II) Methodology:

The model dynamically estimates the percentage increase in COVID-19hospitalizations related to eviction and the costs of those hospitalizations. Thepublic and social costs are based on four inputs:

• The urban population of each of the fifty states;• The absolute value of the decrease in infection rate due to

intervention;• The hospitalization rate as a percentage of the infection rate; and• Public and social costs per hospitalization.The social and public costs related to COVID-19 infection can be calculated

by multiplying the Urban Population by Intervention Effect on Infections by theHospitalization Rate by the Cost per Hospitalization.

To calculate the decrease in infection rate caused by the intervention, itbegins by calculating the decrease in eviction rate. The model starts with theurban population263 because the source study that estimates the effect of a changein eviction rate on COVID-19 transmission only simulated urban spread.264 Itassumes a baseline infection rate of 23%.265 This baseline infection rate matches

260. Jowers et al., supra note 71.

261. Justin Sheen et al., The Effect of Eviction Moratoria on the Transmission of SARS-CoV-

2, at 1 (Jan. 19, 2021) (unpublished manuscript) (on file with the Indiana Health Law Review).

262. Benfer et al., supra note 36, at 11.

263. Urban Area Facts, U.S. CENSUS BUREAU, https://www.census.gov/programs-surveys/

geography/guidance/geo-areas/urban-rural/ua-facts.html [https://perma.cc/8A9M-UD7J] (last

visited Feb. 22, 2021).

264. Sheen et al., supra note 261, at 4.

265. Youyang Gu, Path to Herd Immunity Normality: 2020 Outlook of COVID-19 in the US,

COVID-19 PROJECTIONS, https://covid19-projections.com/path-to-herd-immunity/ [https://perma.

cc/2FQB-K7HY] (last updated Feb. 21, 2021) (using estimates from January 1, 2021).


353

the low end of Justin Sheen et al.’s Strong Lockdown and Second WaveScenario.266 Leveraging the University of Pennsylvania study,267 and annualizedeviction risk projections,268 and considering the baseline monthly displacementrate leads to a baseline of approximately 2% without intervention. This baselineunder the current infection rate would suggest a 10.8 percentage point increasein infections over a six-month period associated with displacement absentintervention.269 It dynamically estimates the delta in eviction rate based on thedifference between the baseline rate and the rate after intervention.

Based on the delta in eviction rate, it estimates the decrease in infection rate.The model again leverages the simulated relationship between eviction rate andinfection rate270 to establish the effect of the intervention on the infection rate.

In order to determine the percentage of individuals who would not otherwiseavoid hospitalization but for the intervention, the model multiplies the estimateddecrease in infection rate271 by the estimated hospitalization rate by state.272 Themean hospitalization rate was used; where data was not reported, the model usedthe national average.273

The average cost of a COVID-19 hospitalization was $23,489.274 Apercentage of total cases were considered to be “public costs.” Assuming constantrates of public expenditure, 41% of these costs are considered public costs basedon the percentage of COVID cases that have been covered by Medicaid,Medicare, and for uninsured individuals.275

4. Additional Limitations

This analysis is further limited by insufficient data on the outcomes ofeviction and displacement. It, therefore, is likely to understate the costs avoidedby evictions avoided. This is particularly true given the limited academic researchon outcomes for those who do not go through the formal legal process of evictionand/or those who do not end up homeless. Indeed, it was difficult to estimatecommensurate mental health, chronic health, foster care, and criminal costs forthe estimated 78% of people facing eviction who likely do not ultimately


267. Id.

268. See supra Part II.B (describing eviction risk).


270. Id.

271. US Currently Hospitalized, COVID TRACKING PROJECT, https://covidtracking.com/

data/charts/us-currently-hospitalized [https://perma.cc/8XJ6-V9KZ] (last visited Apr. 7, 2021).

272. Id.

273. Id.

274. Chris Sloan et al., COVID-19 Hospitalizations Projected to Cost Up to $17B in US in

2020, AVALERE (June 19, 2020), https://avalere.com/insights/covid-19-hospitalizations-projected-

to-cost-up-to-17b-in-us-in-2020 [https://perma.cc/G6F3-ANB7].

275. Id.


experience homelessness. The model does not incorporate any estimates to thateffect. Simply put, the costs are likely to be much greater. Additionally, themodel does not include a significant number of public costs that would beavoided, including education costs, juvenile justice costs, and welfare systemcosts associated with homelessness and displacement. It also does not considerfuture economic consequences that affect tenants’ credit and ability to rent newunits. It does not measure the costs of family and community instability if thetsunami of evictions were allowed to hit uninterrupted.